RBA Annual Conference – 1992 Inflation and Disinflation in Australia: 1950–91 Glenn Stevens^[*]

1. Introduction

There have been several surveys of inflation in Australia in the past two decades. Simkin (1972) looked at inflation from the mid-1950s to the late 1960s. Pitchford (1977) contributed to Krause and Salant's (1977) volume on world-wide inflation, with an analysis which extended into the early 1970s; Hagger (1978) contributed to Gruen's (1978) survey volume of Australian Economics. Two years ago (at a previous Reserve Bank of Australia conference) Carmichael (1990) presented a survey on experience in the 1980s.

All of the papers noted above examined inflation in the context of either upward pressure on inflation, or continued inflation at rates which were relatively high. More recently, of course, inflation has fallen sharply, to a rate lower than when any of the above wrote. At this stage, it would appear that Australian experience with inflation in the 1990s will be quite different to the preceding two decades.

With the rate of inflation at levels not seen for a generation, there seems to be a case for a paper which examines the inflation record over the whole period from the late 1940s to 1991. This paper attempts such an exercise in section 2. It examines in particular several important episodes of inflation and disinflation: the early 1950s; the early 1960s; the early and mid-1970s; the early 1980s; and the period from 1985 to 1991. The aim is to delineate the sources of important inflationary shocks, both domestic and foreign, and the mechanisms through which they worked. In addition, there has to be some discussion, naturally, of the policy responses to those shocks, which included monetary policy, incomes policy and, particularly in some of the earlier episodes, fiscal policy.

While the first part of the paper focuses on the mechanisms at work in episodes of big changes in inflation, emphasising business cycle fluctuations, inflation is not, of course, just a business cycle phenomenon. Monetary policy is the most important determinant of long-run inflation performance, and accordingly section 3 of the paper talks about the role of monetary policy explicitly, in the context of the historical record.

Section 4 of the paper talks about the costs of disinflation. This seems especially apt, since Australia is presently in a period of major disinflation, and since there is on-going debate over the appropriate inflation objective for monetary policy, in which some positions would require further disinflation. No discussion is undertaken here of the costs of living with inflation; a separate paper by McTaggart addresses these issues.

The paper does not attempt a comprehensive treatment of the theory of inflation. That is also the subject of another paper at this conference. Since the sources and nature of shocks have varied over time, and the structure of the Australian economy has changed (in some respects almost out of sight), outlining a unified but sufficiently general theoretical paradigm, and then trying to test it statistically, would be a fairly major undertaking. Generally speaking, however, the paper fits into a fairly mainstream framework. The key ideas are: there is a short-run Phillips curve which describes the price-output dynamics of the economy (though knowledge of its properties is quite imprecise); the economy is subject to various shocks, some of which will shift the short-run trade-off; there is no negatively-sloped long-run trade-off. The paper contains pieces of statistical work which seek to bring light to the issues, but they are not part of a completely specified model, and there is certainly no claim that results are in any sense definitive.

2. The History of Inflation and Disinflation in Australia

(a) Background

Figure 1 shows the rate of change of two key price measures for Australia over the period 1950–1991: the deflator for private consumption and the deflator for GDP from the national accounts.^[1] For the most part, the focus of this paper will be on the consumption price series. It is worth noting, however, that while the overall trends are similar, terms-of-trade fluctuations can drive a wedge between consumer prices and GDP prices over short periods.

Both long-term and short-term features are worthy of note. The short-term feature is the pronounced cyclical nature of inflation rates: peaks in the early 1950s, with lesser peaks around 1957 and 1961; peaks at successively higher levels in 1972, and 1974; peaks in 1982 and 1986 at successively lower levels. Between the peaks, inflation fell away quite noticeably, reaching a cyclical minimum usually within a year to two years of the peak. The only period where inflation was apparently stable for any length of time was between 1963 and 1968.

The long-term features are the downward trend in the 1950s and early 1960s, with each cyclical peak successively lower, the stability in the mid-1960s, a pronounced upward trend between 1968 and the mid-1970s, and (from the vantage point of 1992) a clear downward trend in the period since 1982.

The cyclical pattern suggests that big changes in inflation over short periods are essentially business cycle events, or can be attributed to specific shocks. In fact, many of Australia's business cycles and the corresponding inflation fluctuations have been driven by foreign forces, but domestic shocks have also been important on several occasions, frequently amplifying the foreign impulses. These issues are taken up in general terms in section (b) below.

(b) Sources of Shocks

(i) Foreign Shocks

Australia has always been an open economy (or in the parlance of earlier decades, a ‘dependent’ economy), and foreign economic disturbances have always played a major part in the ebbs and flows of the business cycle. So it is natural to look for foreign influences on the rate of inflation. A variety of theoretical models give the result that for a small country, foreign inflation will be fully imported, in the long run, under a regime of fixed exchange rates, which operated in Australia for the whole of the twentieth century until the early 1970s.^[2]The mechanisms involved, or more particularly the mechanisms emphasised and the assumptions employed, differ between models to some extent. There is not space here to go through all the possible channels in detail, but the key idea is that the small country ultimately has no independent monetary policy under a fixed exchange rate. OECD (1973), Argy and Carmichael (1978) and Salant (1977) (among a host of others) give detailed treatments of the various approaches.^{[3] [4]}

That foreign forces have been a powerful influence is suggested strongly by Figure 2, which shows annual data for CPI inflation for Australia, and a simple arithmetic average of inflation in the United Kingdom (Australia's dominant trading partner until World War II) and the United States (the major partner for much of the post-Warera)^[5] over the 20th century. From the early 1960s, the OECD series for inflation in the G7 countries is used.

The high-frequency fluctuations are not all that closely correlated, in the early period particularly. But there is a clear and strong relationship between the low-frequency changes – the really big movements in inflation which last for several years. The high inflation during and immediately after World War I, deflation in the 1930s, high inflation in World War II and the early 1950s, the upward drift in the late 1960s, the big rise in inflation in the mid-1970s – all these are common to the various series.

Figure 3 shows changes in import, export and consumer prices since the early 1950s. In every case of a major fluctuation in Australian inflation up until the mid-1970s, one can point to changes in either import or export prices (or both) which signify an international event germane to the Australian outcome. After that, changes in the exchange rate became increasingly important (though these often signified international shocks as well), and Australia's inflation rate tended to diverge from the G7 average.

For Australia, it is worth emphasising that one particularly important sort of shock is a large and sudden change in the terms of trade, and in particular a big change in export prices. In the event of a large rise in export prices under a fixed exchange rate, exporters enjoy a boost to their incomes, which shows up initially as an increase in their net financial position (i.e. their bank deposit) with a corresponding asset held by the central bank in terms of foreign exchange. This additional income is likely to generate multiplier effects through the economy, adding temporarily to demand and output and permanently to prices, with the process validated by an expansion in the money stock.

Terms of trade shocks tend, however, to be only temporary. While there is, as is well known, a long-term downward trend in Australia's terms of trade, this trend is quite gentle compared with the size of the cyclical fluctuations around the trend. This means that at some point, the inflationary dynamics discussed above go into reverse: income and spending will fall back, output will temporarily contract, and prices will fall again.

Under floating exchange rates, there is in principle no reason why domestic inflation should be tied to foreign inflation. The economy can be insulated from nominal shocks, since the central bank is in control of domestic financial conditions. Foreign monetary impulses affect the exchange rate. Changes in the world general price level would only be expected to be reflected in the domestic price level to the extent that the central bank was prepared to accommodate this outcome. If policy was completely non-accommodative, these shocks should find their long-run reflection in the exchange rate. Domestic forces, including impulses caused by monetary policy, will also affect the exchange rate.

It is more difficult to say what should happen in the case of real shocks, such as a change in the terms of trade, under floating exchange rates. A permanent rise in export prices should lead to a rise in the exchange rate. Incomes are permanently higher, so consumption will rise, raising demand for domestically-produced goods. Output will rise relative to potential, putting upward pressure on prices. A non-accommodative monetary policy will ensure that interest rates rise, which both reduces domestic demand and switches some of it towards foreign-produced goods by increasing the exchange rate. A temporary rise in export prices might be expected to leave most variables unchanged: permanent income has not changed, so consumption should not change, and there is no need for any relative prices to change. In practice, since it is not known with certainty that a shock is temporary, some adjustment is likely to take place. (Even terms of trade shifts which were later reversed have usually persisted for a year or more.) There is thus some reason to expect that the exchange rate might have a pro-cyclical pattern. Exchange rates may also overshoot, and so the pattern of prices over the cycle might be more muted.

An additional complication may arise if capital flows can drive exchange rates independently of the real economy over short and even quite long periods. Exchange rate changes might then affect the domestic price level and the economy generally, and observers might have difficulty explaining exchange rate and price level movements in terms of standard theory. Exchange] rate theory has struggled to explain some aspects of the floating rate experience in a number of countries.^[6]

(ii) Domestic Shocks

This emphasis on the role of foreign shocks should not obscure the importance on key occasions of domestically-generated shocks, nor should it exonerate domestic policies from responsibility for the medium-term rate of inflation, especially in periods where the exchange rate has been more flexible.

Demand shocks may, of course, originate from policy settings. Supply-side shocks, especially in the labour market, are also potential candidates for discussion. Much of the development of wages has been the endogenous response of the system to forces acting on it, but wage shocks, while far from the whole story, were important in a couple of key episodes. A rise in wages enforced via the legalistic framework of the arbitration system can be seen as analogous to an oil-price shock: it shifts the aggregate supply function in an unfavourable direction. What rate of inflation emerges depends on how accommodative the settings of demand management policies are.

Maintaining a fixed exchange rate constitutes non-accommodative policy, and any attempt to pass on a rise in labour costs into prices will result in a contractionary effect on the economy through a rise in interest rates and a loss of international competitiveness, thus setting in train corrective forces. Under a floating or at least variable exchange rate, a more accommodative stance of policy would be feasible, allowing some decline in the exchange rate and some acceleration in the pace of monetary growth. In all of this, incomes policies, if they are in place and are effective, will also need to be taken into account.

Figure 4 shows a comparison of growth in consumer prices and average weekly earnings. There is close correlation between the two for most of the period. On average, real wages rose. A particularly long period of rising real wages was notable towards the end of the 1960s and in the early and mid-1970s, with a sharp rise right at the end of this period. There was another sharp rise in real wages in the early 1980s. Both of these periods are generally regarded as being characterised by supply-side shocks to wages. After 1982, the relationship between wages and prices became much less clear, and the fall in real wages between 1985 and 1988 was just about unprecedented in the post-War period. The tendency for turning points in wages to lead those of prices was also much less evident in the later 1980s. These simple facts highlight the need for a discussion of the role of incomes policies in that period, which is taken up later.

Figure 5 shows the time path of the ratio of real GDP to trend.^[7] It seems clear that fluctuations in inflation over several years are associated with business cycle fluctuations. This suggests that reductions in inflation involved a period of below-capacity output. It also suggests that to understand the history of inflation, it is worth trying to consider the historical episodes of inflation and disinflation individually. This is taken up below.

(c) Chronology of the Key Episodes

(i) The 1950s: Foreign Shocks

The first major episode was the ‘Korean War boom’ in the early 1950s. The shift from a war-time footing to peace-time priorities in the latter 1940s had posed some adjustment problems. A large scale immigration program was being developed, which added pressure for investment. At the same time, Commonwealth Government control overprices and wages, a war-time expedient, was relinquished. Fiscal and monetary policies, according to Waterman (1972), were accommodative, out of ‘lingering fear of recession’. World inflation was high at the time, so there was some imported component to high Australian inflation of this period. The terms of trade were rising, and the rural sector enjoyed good conditions. Industrial disputes increased, and there was upward pressure on labour costs.

The Korean War broke out in June 1950. By March 1951, wool prices had risen by about 250 per cent from their mid-1950 level. Wool was Australia's major export, and the aggregate terms of trade rose by 50 per cent in 1950/51; export receipts surged 60 per cent. The current account of the balance of payments moved sharply into surplus, and in the absence of any notable offsetting private capital flows, the level of the central bank's foreign exchange reserves rose by over 40 per cent, and its total balance sheet by 25 per cent. M3 (currency and bank deposits) rose by 19 per cent, as exporters' receipts were turned into Australian currency and banked. The increased incomes and money balances fed into higher domestic spending and output. Real GDP expanded by 17 per cent over two years, 1949/50 and 1950/51. Real domestic demand expanded even faster. Coming on top of already high inflation, this stimulus pushed up inflation sharply and in the year to December 1951, the consumer price index rose by 25 per cent. Wages were at this stage still being adjusted annually for movements in the CPI; they rose sharply in real terms at the height of the boom, as pressure on labour markets was intense. (In the middle of 1951, the total number of people receiving unemployment benefit in Australia was a there 600.)

Government policy response to this situation is discussed by Waterman (1972). Proposals for a revaluation of the Australian pound, or a float along Canadian lines, were not favoured. The Commonwealth Bank (at that time the central bank) issued directives about credit growth to the banks, and made very large calls to ‘special account’ (the precursor to SRDs/NCDs). There was a modest rise in interest rates. The Government raised some taxes, including imposing prepayment of taxes on wool receipts, but at the same time increased expenditure sharply, so the 1950/51 budget was not really contractionary.

By the middle of 1951, however, wool prices had collapsed, falling by half in the space of three months as a result of substantial cutbacks in US purchases related to the war. These external events were reinforced by the effects of the 1951/52 ‘horror budget’, which increased taxes sharply, and pushed the Commonwealth Government's financial balance towards surplus by about 3 per cent of GDP. In addition, imports had grown strongly, a lagged result of the build-up in incomes and spending.

The current account went sharply into deficit in 1951/52 as a result, and all the inflationary forces outlined above went into reverse: the central bank's holdings of reserves fell by half; M3 fell by about 1 per cent; spending, output and demand for labour contracted sharply. Waterman (1972) dates the downturn as beginning in late 1951. Its full effect in the year average figures available in the national income accounts show up in the 1952/53 financial year, when real domestic demand was 13 per cent lower, and real GDP almost 5 per cent lower, than in the preceding year. There was a rapid fall in consumer price inflation, from 25 per cent at the end of 1951, to zero by late 1954. Growth in wages fell back too, from close to 30 per cent to around 5 per cent.

(ii) The Early 1960s: a Domestically Generated Business Cycle

There was another, more minor, cycle in activity and inflation in 1956–58. This can be regarded, and was certainly seen at the time, as externally generated in a qualitatively similar way to the early 1950s episode above. I do not consider it here, though the experience is included in some comparisons of disinflations made in Section 4 of the paper.

The cycle in the period 1959–62 is of more interest, since domestic forces played a bigger role, certainly in the down-swing of the cycle. The economy had largely recovered from the growth recession of 1956–58 by 1959. Robust world growth and a stable terms of trade helped, but the main source of growth was domestic demand, which was expanding quite strongly. The labour market was showing signs of excess demand, with wage rises running considerably faster than prices.

The rate of consumer price inflation was around 2 per cent, but moved up noticeably in 1960, reaching a peak in the year to September 1960 of around 5 per cent. Policy makers were concerned about the general inflationary climate; not just the rise in consumer prices but also the rate of increase in property and share prices. The share price index had risen by two thirds between the end of 1957 and the middle of 1960. Residential property prices had risen strongly right through the second half of the 1950s, under the influence of rapid expansion of housing demand.^[8]

Policy responded to this situation initially by repeatedly asking banks to slow their lending, and by easing licensing restrictions on imports. The latter set in train deflationary forces by allowing excess demand to be met from abroad. It also set up, through the effect on banks' liquidity, the conditions for a sharp tightening of credit conditions. In November 1960, the Government also tightened fiscal policy, raising sales tax on motor vehicles, among other measures. The final leg was a rise in regulated bank interest rates.

The effect was to dampen confidence quite sharply. Looking at the figures, the drain on foreign reserves through higher imports was offset by private capital inflow (and a drawing on the IMF in April 1961), but there was a sharp decline in the ease with which credit could be obtained around the end of 1960. The economy weakened quickly in the first half of 1961, with real GDP falling by around 3 per cent between its peak in the September quarter 1960 and the low point a year later, which was about 4–5 per cent below trend. The 1961 recession thus ranks as one of the more severe downturns of the post-War period. The contraction in credit and activity quickly reduced inflation. Consumer price inflation ran at around 5 per cent in the year to March 1961, but prices were virtually stationary for the next two years.

This episode was characterised, as noted above, by a much bigger role for apparently domestic factors. There were international aspects, but these were nowhere near as important as they had been in the episodes in the 1950s, or would be in the 1970s. Testimony to this is the fact that Australia's rate of inflation behaved quite differently in the first half of the 1960s to the average of the big countries.

Even so, developments abroad still loomed large in the thinking of Australian economists. Reading through the succession of articles entitled ‘The Australian Economy’ in the Economic Record in this period, one is struck by the way in which balance of payments considerations were seen as driving business cycles and constraining the policy responses, even though the 1959–62 cycle seemed domestically driven for the most part. As it turned out, the steady growth of the world economy in the 1960s provided a relatively benign environment for Australia. No sooner had Lydall (1962) coined the term ‘brooding pessimism’ as a description of the widely-held view of economists about Australia's balance of payments, than the constraint imposed on the economy and on policy by balance of payments concerns seemed to lessen. The period from 1962–68 stands out as the period of least foreign disturbances in the post-War era. Export price fluctuations were small compared with the 1950s or the 1970s and 1980s; import prices were remarkably stable; and the terms of trade tended if anything to rise. As a result, Australia enjoyed a good combination of low and stable inflation, strong output growth and very low unemployment. It almost seemed reasonable to think about inflation in terms that were akin to a closed economy. Simkin, writing in 1972, felt able to concentrate on a discussion of ‘cost push’ and ‘demand-pull’ explanations in his analysis of the period from 1953 to 1971, paying little attention to foreign factors. Writers coming only a short time after him, by contrast, were obliged to reopen the whole question of ‘imported inflation’.

(iii) The Early and Mid-1970s: Foreign and Domestic Forces

Part of the story of the third major episode – the rise in inflation in the first half of the 1970s – can be told in similar, though not identical terms to the early 1950s. The general history of this period from an international perspective is well known. Key elements were a build up in world liquid assets; an upward drift in inflation and inflationary expectations in the late 1960s and the first couple of years of the 1970s, against which the policies put in place seemed ineffective; an increase in mobility of financial capital; recurring speculation against the US dollar, which was the main reserve asset under the Bretton-Woods exchange-rate arrangements; a boom in commodity prices; and especially a quadrupling of the US dollar price of oil in 1973.

For the world economy, the commodity price shock on this occasion was much bigger than in the early 1950s. For Australia, there was a large rise in the terms of trade, which boosted exporters' incomes, and set off many of the same sorts of processes as had been seen before: rises in money, spending, output and wages and prices. This effect was much smaller than in the earlier episode: the terms of trade rose only about half as far. But even before this happened, the international instability of the early 1970s had affected Australia via a channel which had become much more important since the episode of two decades earlier: the mobility of private financial capital. Capital inflows in 1971 and 1972 were large, as the Australian dollar was perceived as undervalued. In the latter year, the world commodity price boom also affected export receipts, and pushed the current account into surplus. Both these factors pushed up the Reserve Bank's holdings of foreign reserves, and monetary growth accelerated sharply.

Policy responses to this involved a revaluation of the $A in December 1972, tightening of controls over capital inflows (including the Variable Deposit Requirement), and a further revaluation in September 1973. These measures succeeded in choking off the foreign capital inflows, which turned to outflows in the latter part of 1973. At the same time, the boom set off by the expansionary forces had already prepared the path for its own reversal – expansion of domestic incomes and spending had pushed up imports, and the current account surplus evaporated in the first half of 1974. These events combined with the usual seasonal pressures in the financial markets to create a severe liquidity squeeze in the June quarter of that year. Foreign capital inflow helped to equilibrate the system, but only when interest rates on market securities had reached unprecedented levels. Monetary growth fell rapidly to zero or less, and a ‘credit crunch’ was imposed on the economy.

This ‘crunch’ was, of course, only short-lived. As domestic spending began to contract, imports fell off. The exchange rate was devalued in September 1974, and capital controls were eased, which ensured that capital inflow could be associated with lower domestic interest rates.

The initial pick-up in inflation in the early 1970s, and the Korean War boom and bust, were both good examples of the conventional view of the transmission of disturbances under a fixed exchange rate. Foreign prices influenced the domestic price level both directly, and through effects on money, income and expenditure, all of which rose and fell more or less as would be expected. Jonson's (1973) results using the first generation Reserve Bank model suggest that 60 per cent of the inflation over the four years to 1973(2) could be attributed to ‘imported’ inflation. Pitchford (1977) gives a prominent role to external shocks in initiating the inflationary process of 1972–74, as do Argy and Carmichael (1978).

External shocks continued to be important in the Australian experience after this. Australia did share, to some extent, in the world disinflation of the second half of the 1970s, and in the re-emergence of inflation at the end of that decade and the subsequent disinflation of the early 1980s, albeit somewhat belatedly in each case.

But Australia began to move more independently of the rest of the world. At this point, domestic shocks and policy settings became much more important. Fiscal policy was strongly expansionary. Wage developments also proved to be a key factor. An increasing degree of exchange rate flexibility went hand in hand with this. The case for floating exchange rates, which had been so compelling to other countries in the mid-1970s environment of extreme international monetary instability, was embraced only slowly by Australia, with ‘the triumvirate’ adjusting the rate daily from November 1976, and the full float only coming nearly a decade later. But even the limited flexibility that crept into exchange rate regime in the mid-1970s meant that the old mechanism which had linked Australia to the world rate of inflation quite closely, helping to push inflation up in the booms but relentlessly deflating the economy in the other phase of the cycle, now became slightly more fluid.

No consideration of the mid-1970s would be complete without a discussion of wages. One immediate problem, of course, is that it is not straightforward to isolate the exogenous elements of wage changes from the endogenous responses to other forces, particularly in an environment of rapidly rising prices and wages and changing inflationary expectations: which came first, price increases or wage increases?

This discussion over wages – which has taken place in many countries – has been further complicated in the Australian case by the role of the Australian Conciliation and Arbitration Commission. The Commission's legal powers are such that it has clearly been in a position to exert some exogenous influence on minimum wages, and the very early history of the Commission was one of doing just that, with a goal of social reform. Its more recent behaviour, and its influence over average wages actually paid, has been a subject for more debate – does the Commission exert an independent effect on actual earnings, or has it responded to the state of the labour market, in a way which might even roughly mimic a market-determined outcome? Even by the mid-1970s, this debate had been running for some time – and it was still running in the late 1980s.

The exact nature of the approach adopted by the Commission has varied over the post-War period; but it was usually reluctant to see its primary role as limited to, or sometimes even connected with, macro-economic policy efforts to control inflation. Except in unusual circumstances, it responded to price inflation by increasing wages. Automatic annual cost of living increases in the ‘basic wage’ were awarded until 1953. They were then abolished on the grounds of insufficient ‘capacity to pay’, but reinstated briefly in the early 1960s. Quarterly indexation of various forms was practiced from 1975 to 1981. Non-indexation increases based on the ‘capacity to pay’ were also made from time to time, and the Commission considered cases for changes to ‘margins’ in key industries.

The Commission could not, furthermore, really prevent market forces from increasing over-award payments at some times, and some of the key changes in the Commission's modus operandi have been motivated by a desire to bring wage setting back into its own orbit. Agreements of this nature eventually led the Commission in the late 1960s to abolish the idea of the basic wage and margins, and focus more on total wages, in an effort to regain control over the wage setting process. Forces of a similar nature in the early 1980s led the Commission into new territory again.

While some early econometric studies of wage behaviour (e.g. Higgins (1972)) treated award wages as exogenous, and as the key explanatory variable in the determination of average earnings (with traditional labour market pressure variables usually playing a role in this second stage as well), Jonson, Maher and Thompson (1974) concluded that award wages responded to actual and expected future inflation. Hancock (1971), who quotes members of the Commission, embodies a similar view. The very vigorous Australian debate over the nature of the short-run Phillips curve waged mainly in the pages of Australian Economic Papers adopts a similar approach, by modelling wages as responding to inflationary expectations and a variable connected with pressure on labour resources.

The idea of an exogenous role for the Commission remains strong, however. It is seen in Boehm (1984), for example, who finds evidence of Granger causality from wages to prices. The strong focus on the use of incomes policy in Australia in the 1980s in particular can also be seen as presupposing an important role for the Commission (though it is also dependent for working success on strong centralised decision making within the trade unions).

With that background, we can now examine the 1970s from a wage perspective. By the end of the 1960s, the Australian economy had been growing at an average annual rate of 6 per cent for eight years. Unemployment had been consistently below 2 per cent in that period. The only peace-time precedent for that in the twentieth century had been in the immediate post-War years, which had been associated with a rise in inflation to over 9 per cent, and then the acceleration into the Korean War boom. Inflation, which between the end of 1953 and the end of 1963 had cycled between zero and 5 or 6 per cent, with an average of about 2 per cent, was much steadier, and averaged about 3½ per cent between the end of 1963 and the end of 1970. Business profitability was high, and labour unrest picked up as the 1960s drew to a close. The balance of payments was now being underpinned by the emergence of new possibilities for mineral exports. Mineral discoveries were also at the centre of a speculative boom in the share market. It is hard not to conclude that this long expansion, with domestic economic management temporarily unconstrained by balance of payments considerations, contributed to a gradual pick-up in inflation, and helped to entrench the idea that continuing inflation was a fact of life. The minor downturn in economic activity in 1971 did little to dampen this enthusiasm, and had only a temporary and small effect on actual inflation. The Arbitration Commission increasingly struggled to retain control of the wage-setting process, and real wages continued rising steadily.

To this point, it is possible to conclude that excess demand was the dominant influence. But there followed developments which were more clearly of the nature of a shift in the supply of labour. The Whitlam government, for example, was supportive of large pay rises before the Arbitration Commission, and used its influence over the Commonwealth Public Service to spearhead the push for improvement in conditions – in terms of pay rates, amount of paid leave, etc. – so as to effect a redistribution of national income towards labour. There were also, importantly, the decisions to enforce equal pay for women – the ‘equal pay for equal work’ decision in 1969 and the ‘equal pay for work of equal value’ decision in 1972. The thrust of these decisions was to raise the minimum wage for women from 75 per cent of that for males (where it had been set in an earlier decision) to 100 per cent, with the adjustment to be phased in over several years. (Part of the Government's wages push was to introduce equal pay in the Public Service immediately.)

The moral and social aspects of this issue aside, it seems clear that moves to increase females' pay to be commensurate with males' resulted in a substantial rise in aggregate labour costs. Pitchford (1977) carefully compares wage increases for males and females between 1969 and 1974. His analysis shows that between the end of 1969 and the end of 1974, male average hourly wage rates rose by 98 per cent, while the corresponding rise for females was 142 per cent.^[9] Compared to a situation where all wages had risen at the male rate, and given that women comprised about a third of the labour force at the time, the more rapid rise in female wages must have added substantially to aggregate labour costs. Rises in real wages had been seen in the upward phase of previous cycles, but the rise in real wages over several years up to 1974 was unprecedented.

In summary, it is hard to go past Pitchford's (1977) conclusion that (i) the big 1970s inflation was set off largely by foreign factors, operating through similar sorts of mechanisms as in the early 1950s; but that (ii) domestic factors added to the effects of these foreign shocks, mainly through operating to increase wage costs. Thus Australia's inflation rate rose by more than did those of most of the major countries, not (as had been the case in the 1950s) because of the extreme nature of the terms of trade shock and the effects it set off, but because domestic factors initially reinforced the foreign shock.

(iv) 1975–80: Disinflation and Wage Indexation

Efforts at disinflation in the second half of the 1970s relied on both monetary and wages policies. On the monetary front, Australia joined the international trend to announcing monetary targets in early 1976. In that year, and in 1976/77 and 1977/78, targets for growth in M3 were set, and met, at successively lower growth rates.

In the wages arena, the most important issue of the time was indexation. The Arbitration Commission adopted the practice of holding quarterly wage hearings in April 1975, and began awarding wage increases based on the CPI increase of the preceding quarter.^[10] There was considerable discussion about whether indexation was itself likely to contribute to higher or lower inflation; Nieuwenhuysen and Sloan (1978) give an account of this in the longer-term Australian setting.

The Whitlam Government's support for indexation in April 1975 was on the basis that it would restrain wage increases (Nieuwenhuysen and Sloan (1978)). Proponents of indexation argued that it would reduce the costs of disinflation, by linking wage increases to actual inflation (which would be falling) rather than to expectations of inflation dominated by recent very high figures. This depended a great deal on how inflationary expectations were formed,^[11] but assuming largely adaptive expectations, and given that the peak rate of inflation of about 19 per cent was almost certainly a once-off result owing a lot to the rise in import prices which would not continue, tying future wage increases to the slowing rate of actual inflation was preferable to an alternative of having wages increasing in line with expectations of inflation, which would be heavily influenced by the extreme experiences at the peak. An eyeball examination of the data in Figure 4 certainly does not suggest that indexation added to inflation, since inflation began falling at about the same time as indexation was introduced (though it says nothing about how fast inflation might have come down under some other policy). The policy tied down real wage growth to roughly zero for about five years, moreover, an outcome unprecedented in the post-War period.

The validity of indexation as a disinflationary tool depended, however, on two assumptions. One was that other policies were set in a disinflationary mode. The second was that the shocks involved were purely nominal, not real in nature. A potential flaw existed in the first case in that, after initial successes, Australia's record in pursuing monetary targets as an effective counter-inflationary weapon was not particularly noteworthy. In 1978/79, monetary growth exceeded the target by a substantial margin. Later targets were set higher, and no real attempt was made in subsequent years to push monetary growth onto a declining path.^[12] In fact, with the regulations governing key financial prices, monetary policy was incapable of offering much independent force in the fight against inflation at this time. Fiscal policy was also such that substantial budget deficits still existed until very late in the decade, which posed more or less continual funding problems and impinged on monetary growth.

The second problem with indexation was that the level of real wages which was locked in was too high, initially at least, by any standard. It is true that real wages rose sharply in all countries in the mid-1970s but the rise in real wages relative to productivity in Australia stands out quite distinctly in international comparisons (see OECD Economic Outlook, July 1977, pp 62–63). This constituted a large and adverse supply shock. Under such circumstances, when relative prices needed to adjust, indexation was bound to hamper adjustment. Indeed one of the major arguments mounted by the Fraser Government against awarding wage increases in the latter 1970s was the existence of the ‘real wage overhang’ – the extent to which cumulative real wage increases in earlier years had outstripped gains in productivity.

It seems appropriate to conclude, as does Phipps (1981), that while indexation may have helped to adjust wage inflation to falling price inflation initially (i.e. it helped to adjust to a nominal shock), it held back adjustment to the real shock and thereby damaged prospects for output and employment growth in the latter part of the 1970s.

It can also be observed that the prevalence of indexation of wages does not sit well with the emerging tendency to vary the exchange rate in response to balance of payments pressures, such as the decision to devalue in late 1976. The net fall in the exchange rate over the second half of the 1970s amounted to about 17 per cent from its 1970 value, and 30 per cent from its peak of early 1973. Partly as a result of this, the rate of change of import prices in $A terms between the end of 1975 and the end of 1980 averaged 13 per cent per annum, and never fell below 5 per cent. This made the task of reducing inflation more difficult, even though the Arbitration Commission sought to discount wage rises for the measured effects of exchange rate depreciations on some occasions.

In short, Australia managed to join the international disinflation in 1975, and inflation fell away quickly from nearly 20 per cent at the peak to about 10 per cent by 1977. But further success proved difficult. A slowing in the economy in 1977/78 helped to push inflation down temporarily, but the lowest recorded annual rate of inflation in the second half of the 1970s was just over 8 per cent in 1978. Thus Australia went into a period of stronger growth, and rising oil prices, which ultimately became the second OPEC shock, without having managed to reduce inflation below the high single-digit range.

(v) 1979–84: OPEC II and the ‘Resources Boom’

The second big rise in oil prices began in mid-1978. By mid-1979, the US dollar price of oil had doubled. Inflation in the OECD economies rose from early 1979 reflecting this, reaching a peak in the middle of 1980. For the G7 group as a whole, the peak was at a lower rate than in 1974, mainly reflecting vastly improved performance for Japan; US inflation, by contrast, peaked at a much higher rate in the early 1980s than in the mid-1970s.

The rate of change in the G7 export price deflator rose to about 11 per cent through calendar 1979; for Australia's basket of imports, the rise was closer to 20 per cent in foreign currency terms over the same period. The effect of this in $A terms was lessened only slightly by an appreciation of the exchange rate – over the course of two and a half years from early 1979, the TWI rose by about 14 per cent, but this came on the back of the substantial depreciation of the preceding few years.

Export prices were also rising quickly again; the export price deflator rose by 23 per cent through 1979. This did not represent much of a gain in the aggregate terms of trade, since import prices were rising at a similar pace. But the conjunction of high world energy prices and Australia's relatively rich endowment of energy resources produced a remarkable sense of euphoria about prospects for the energy sector. In this mood of improved confidence, Australia continued to grow in 1980 and 1981, even while demand and output in the G7 countries stagnated. The labour market showed signs of improvement, after several years in which employment growth had been anaemic. In 1979 and 1980 growth in the number employed outstripped the labour force growth and was sufficient to make very modest inroads into unemployment, though this still remained very high relative to most previous experience. In addition, after several years of closely controlled wage outcomes under the Arbitration Commission's indexation principles, the stronger unions were getting restive, especially in light of the fact that indexation rises in award wages had amounted to only about 80 per cent on average of the relevant CPI increases. (Average weekly earnings had more or less kept pace with inflation, due to ‘work value’ rounds, which were treated separately from indexation increases, and a modest degree of ‘wage drift’.)

An effective campaign in the metal industries, where there was a noticeable shortage of labour, reduced standard working hours, and led to substantially higher wages. A number of other industries also negotiated wage increases outside the guidelines. Under these circumstances, there was growing pressure on the centralised wage setting arrangements by the turn of the decade. The Commission was unable to keep the relevant parties in agreement with its approach. In July 1981, the Commission abandoned the approach that had been in place since 1975, and proceeded to deal with claims on a case-by-case basis. It was an experiment with the market.

But State industrial tribunals continued to award across the board increases related to CPI movements. In addition, the metal trades agreement acted as a pace-maker for other industries in the federal sphere, and the principle of comparative wage justice provided a mechanism to share the gains around. There was as a result a substantial rise in aggregate labour costs. In the three years to the end of 1982, average weekly earnings rose by 45 per cent in nominal terms. Table 1 puts this in context, by comparing with the rise in the three years to the end of 1974.

On balance, the wage shock in 1980–82 does not look to have been as large as in the mid-1970s, but it was still substantial. It re-opened the gap between real wage costs and productivity which had slowly been closing in the latter half of the 1970s. Consumer price inflation had peaked with the rate of import price inflation in mid-1980 and then fallen a little. In the aftermath of the 1982 wage push, it rose again, reaching a peak of about 11 per cent in the second half of 1982. Australia's inflation was now moving in the opposite direction to that of the big countries.

Policy makers focused on the upward pressure on prices, and the widening current account deficit. Unlike in earlier episodes where expanding imports had led quickly to a sharp financial contraction (e.g. in 1952 and 1974), capital inflow funded the deficit for a considerable time (the financing counterpart to the large rise in investment, including by a number of government authorities). This was made possible by the greater upward flexibility of domestic interest rates, which had been rising (in concert with rates around the world, led by the rise in US rates) since the beginning of 1979. The continued inflow kept the banking system from experiencing the same sort of supply-side contraction as in 1974 or the early 1950s; but the rise in market interest rates (90-day bill rates in April 1982 were over 21 per cent) had a profound deflationary effect on the demand for credit and on the economy. Credit growth started to slow in early 1982 as the housing sector began to contract. Unemployment started to rise in mid year, and between mid-1982 and mid-1983 real GDP contracted by about 2½ per cent. The contraction was exacerbated by the effects on the rural sector of a debilitating drought.

As the large rise in unemployment began to take shape, the Government introduced a wage freeze in December 1982. It froze wages for its own employees with an Act of Parliament; a couple of State governments followed suit. The Arbitration Commission acquiesced with the Government's wish to freeze other wages, and the index of award wages was approximately stationary between December 1982 and September 1983. The combination of this and all the other disinflationary forces on the economy produced a sharp fall in inflation, to about 5 per cent by the end of 1984.

It would be some time before this sort of inflation rate would be seen again as the events of the mid-1980s proved to be counterproductive in the short term for efforts to reduce inflation. But the main picture which emerges from the 1979–1982 episode is that a world inflationary shock once again set off inflationary forces in Australia, which was again exacerbated by domestic forces. The shock was somewhat different to the previous two major episodes, which had been terms of trade events for Australia. There was no observable terms of trade rise in 1979–82. But the euphoria over high energy prices can be seen as embodying an assumption that the terms of trade would be higher in future, with the same effects as an actual rise in the terms of trade.

The transmission mechanism had also changed, in that capital mobility, the reluctance to resort to capital controls and a preparedness to let domestic interest rates rise much further meant that the financial restraint came less through a quantity constraint operating on bank's balance sheets – i.e. on the supply of bank money or credit – but through the impact of interest rates on the demand for credit. M3 growth in fact showed no sign that the economy was in the most severe contraction for two decades, continuing to run at over 10 per cent through the recession. The impact of high interest rates could be seen, however, in the part of the financial system which had never been quantity constrained, but always affected by price: the non-bank intermediaries. Broad money and credit – which had for some years been growing faster than M3 – did slow noticeably, and it is no coincidence that the Reserve Bank focused increasingly on these aggregates in its analysis, even though M3 retained pride of place as the object of formal monetary targets. It also seems that to the extent that this tightening in financial conditions was a discretionary act by the authorities (and there was at least some discretion over how much of the movements in foreign interest rates would reflect in domestic financial conditions and how much in the exchange rate, since it was now the practice to vary the latter by reasonably substantial amounts, even if the full float was not yet in operation), it was motivated to a considerable degree by concerns over the deteriorating external position. So, in a sense, policy reaction still provided some of the same ‘discipline’ as the fixed exchange rate had always imposed. The main difference was that the effect was muted in the short term, and took longer to show up in full.

(vi) 1985–1991: Floating Exchange Rates and Inflationary ‘Independence’

After the wages breakout of 1980–82, wage-setting arrangements in Australia reverted to a highly centralised set of arrangements. The wage pause instigated by the Government and acceded to by the Arbitration Commission was the first stage in this. The institutional setting was then bolstered by the Prices and Incomes Accord, an agreement struck between the Australian Labor Party and the Australian Council of Trade Unions in late 1982. With the election of the ALP to government in March 1983, the Accord formed the centre-piece of wage setting for the rest of the decade. In its original form, it called for a return to wage indexation in principle, though there was recognition of the need to restore business profitability in the initial stages of economic recovery. In September 1983, the Arbitration Commission increased award wages by 4.3 per cent, based on recent CPI rises, and set out on a course largely along the lines envisaged by the Accord. There was a further indexation rise of 4.1 per cent in April 1984. This restored some of the loss of real wages which had resulted from the wages pause in 1983.

Actual price inflation was continuing to fall, however. There were two key influences. The first was that the recovery was generating the usual cyclical rises in productivity, helping to keep unit costs down. The second was that lower world inflation was being imported to some extent. Once the effects of the sharp fall in the exchange rate in March 1983 had been absorbed, and indeed reversed during 1984, import prices in $A terms were stable. Then, just at the right time, Medicare was re-introduced, which produced a fall in the March quarter 1984 CPI of 0.4 per cent. This provided a temporary short-circuit for the wage setting process (even though taxes had effectively been raised to fund public health care), and helped to contain inflation.

To this point, the fall in inflation in Australia between late 1982 and the end of 1984 broadly paralleled that in the rest of the world, with a lag. Average inflation in the G7 countries had stabilised at about 4 per cent. In Australia, consumer prices rose by about 5 per cent in underlying terms through 1984.

By 1985, however, growth in the world economy had slowed from the exuberant pace of the early phase of recovery. This is visible in the data for growth in real GDP of the G7, but it is much clearer in the data for industrial production, where growth slowed from a high of about 8 per cent in early 1984 to about 1 per cent in 1986. This was associated with a large fall in oil prices, in which nominal oil prices fell by nearly as much as they had risen in 1978–80. The fall in commodity prices was widespread; the basket of commodity prices most relevant to Australia fell by about a third.^[13] This was the biggest fall in Australia's export prices since the early 1950s. A fall of this magnitude in the earlier decades, under the fixed exchange rate, would have had a pronounced contractionary effect on the Australian economy. On this occasion, a floating exchange rate offered assistance in adjusting to the shock.

A fall in the exchange rate was quite proper under circumstances of a sharp decline in the terms of trade. The decline proved to be large, however, and quite disorderly. The $A came under acute downward pressure in February 1985. There was a further sharp downward movement in April 1985, and periods of intense speculative pressure in November that year and again in mid-1986. By July 1986, at its lowest point, the $A had fallen by more than a third against the TWI from its end-1984 level. Movements of 2–3 per cent on a single day occurred on several occasions.

The reason for this speculative pressure was that as well as declining commodity prices, a number of other factors had also focused attention on the exchange rate. The economy had by early 1985 been recovering strongly from the recession for about six quarters, in which time real GDP had expanded by over 11 per cent, and was back to more or less its trend level (though unemployment was still relatively high). The structural weakness of the balance of payments was being increasingly exposed, and there was uncertainty about the objectives and conduct of monetary policy, in the context of the deregulation in the financial system. There were also reservations about economic policy generally, with a large fiscal deficit and pressures on the Accord, but the overwhelming pre-occupation of the economic policy debate was the balance of payments situation. Compounding this was a degree of political uncertainty. These factors magnified a fall in the exchange rate which was the proper response of the system to the declining terms of trade, into something considerably bigger.

The set of signals coming from the financial markets led to a sharp tightening of macro-economic policies, and monetary policy in particular. Short-term interest rates rose from around 11 per cent in late 1984 to around 16 per cent in mid-1985, then to over 19 per cent at the end of the year.

The fall-out in the real economy was remarkably small. Even though interest rates rose, exporters' incomes were cushioned by the lower exchange rate, and the loss of real income was spread to the community in the form of higher prices for imported products. Real GDP paused in 1986, but did not fall appreciably; the rate of unemployment rose only marginally.

On the other hand, it was inevitable that a fall of this magnitude in the exchange rate would add to the price level. Consumer price inflation picked up from early 1985, coinciding fairly closely with the rise in import prices, and reached a peak of around 9 per cent in the year to December 1986. After that, with the exchange rate stabilising and then rising, it fell back quite quickly to around 7 per cent. The magnitude of the effect on prices of the exchange rate fall itself is uncertain. The ABS measure of increases in the prices of goods ‘wholly or predominantly imported’ contributed about 2½ percentage points to the CPI over the course of about two years. This was only the direct effect, and some other estimates judged the total effect to be considerably higher.^[14]

A fall in the terms of trade had always been deflationary in the past; now, with the exchange rate falling enough to help cushion output, that effect would be lessened or perhaps removed. But there is no reason to think that a terms of trade fall in itself should add to inflation, and so it is necessary to appeal to the other factors present at the time to explain a rise in inflation. It is impossible to disentangle precisely all these effects, but the most plausible assessment would appear to be that the exchange rate declined further than dictated by just the terms of trade, as a result of a number of factors, including a re-assessment of the long-run prospects of the economy even at an unchanged terms of trade, and concerns about macro-economic policies which were seen as being on the accommodative side in an economy which was moving ahead quickly.

To the extent that a once-off adjustment to the real exchange rate was required as part of the adjustment to a structural imbalance, this was likely to show up as a fall in the nominal exchange rate, and as a temporarily higher rate of inflation. A key feature of such an adjustment should be that real wages fall. The Government secured agreement with the ACTU to quarantine at least some of the price rises from the depreciation from wages, and the Arbitration Commission's decision in the May 1986 Wage Case reflected this. It is clear from Figure 4 that real wages did indeed fall in this period. This helped limit the second round price effects of the depreciation, and the extent of the fall out in the labour market, and made for a permanent shift in measures of competitiveness based on labour costs, and in the real effective exchange rate.

In the second half of 1986, the world economy began to accelerate again. Production growth picked up, and commodity prices improved. The $A began to rise at the same time, as the more extreme bouts of pessimism subsided. It was still vulnerable to occasional downward pressure, but these episodes evoked a much stronger response in terms of intervention by the Reserve Bank, and temporarily higher interest rates where necessary (for example, in January 1987), and the general trend in the exchange rate was noticeably upwards.

Initially, monetary policy eased. The economy had been flat through 1986; wages growth was well under control. Sentiment about the $A continued to improve, helped by the rising commodity prices and by signs of fiscal consolidation, as expenditure reviews were beginning to make inroads into the fiscal deficit. With the exchange rate, the main potential threat to prices, looking much stronger, inflation was set to fall. As a result, interest rates declined in the first half of 1987.

This could not persist, however, since the world economy was picking up momentum, and this was flowing through into Australia via the usual mechanism: the terms of trade were recovering strongly. Australian commodity export prices rose by 45 per cent from the low point in mid-1986 to the peak in September 1989. This rise in export prices was comparable with the rise in the early 1970s, and in a striking parallel with earlier decades, wool prices led the way, approximately doubling over the course of calendar 1987. The rise in prices became more widespread as time went on, and the aggregate terms of trade rose by 28 per cent between December 1986 and March 1989.

There was an interregnum in late 1987 in the aftermath of the share market crash, when the deflationary effect of a fall of nearly half in one major class of asset price was overestimated. The pace of growth of demand and output in fact quickened appreciably in 1988, other asset prices continued to rise strongly and in April 1988 monetary policy began to tighten.

The sharp rises in interest rates in 1988 strengthened the upward trend in the $A, which reached a peak in March 1989 about 35 per cent above its lowest0 point in mid-1986. As a result, at the time of maximum inflationary pressure in the Australian economy, when real domestic demand expanded by about 10 per cent (in the year to June 1989), import prices were falling in $A terms. In the year to March 1989, they fell by about 11 per cent.

In this respect, the flexible exchange rate was playing an important role in helping the economy adjust to the sort of shock which had previously always led to a big increase in inflation. The rate of consumer price inflation had come down from over 9 per cent in the immediate aftermath of the depreciation of the exchange rate in 1985 and 1986, to around 7 per cent. The effect of the period of rising terms of trade and general expansionary pressure in the economy in 1988 and 1989 was to check the downward momentum for time, but inflation did not actually rise, in clear contrast to the experience in several previous episodes outlined above. This reflects the flexibility of the exchange rate, and the overall tightness of the monetary policy stance which, by the middle of 1989, had produced very high real interest rates. In the view of most practitioners, monetary policy was very tight by the time overnight interest rates reached 18 per cent and bank prime rates over 20 per cent, even though money and credit figures still showed fast growth.

The success in containing inflation also owed something to the behaviour of wages, which was quite different in the late 1980s compared with the other episodes, when wages responded strongly to the expansionary forces in the economy. Much has been written about the behaviour of wages and the role of the Accord. On one view, it has successfully restrained wage growth, indeed reducing real wages, and thus contributed to strong growth in employment and moderation in inflation in the second half of the 1980s. An alternative view is that this outcome was not due to the Accord itself, but that the centralised arrangements through which the Accord operated merely produced what a functioning labour market would produce anyway. Wage moderation was a feature of many OECD economies, and it has been suggested (Poret (1990)) that this was simply a response to the high unemployment of the period. Pissarides (1989) concluded that a wage bargaining model was quite consistent with Australian data.

The late 1980s episode provides an important piece of information in distinguishing between these two hypotheses. A simple comparison of the situation in 1989 with that in 1982 (the previous occasion when the labour market was about as tight) suggests that the Accord did restrain wages in the latter period, since the acceleration in nominal wage growth was relatively mild, and the magnitude of the expansionary external shock easily as big.

By mid-1990, under the influence of high interest rates, output and spending began to fall back from their extremely high levels. The decline accelerated in the first half of 1991. In total, the contraction of output was similar in magnitude to that in 1981–83, though noticeably smaller than in 1952–53. This was considerably larger than expectations of official (and most other) forecasters, who had noted the absence of a misalignment of labour costs or a major international recession, factors which had in the past been seen as playing an important role in business cycles in Australia. The flip side of this was that inflation came down sharply as well. Having been successfully contained at around 7 per cent in 1989, consumer price inflation fell to around 2 per cent in the year to March 1992, which was the lowest result since the 1960s.

3. The Role of Monetary Policy

The above discussion has focused heavily on business cycle dynamics as the setting for rises and falls in prices and wages. It has sought to elucidate the key shocks which have been associated with big movements in Australian inflation over the past four decades. But it is important to focus on the role of monetary policy in particular for a moment, and also to step back from the short-term cyclical phenomena and look at the longer-term trends, where one presumes that the monetary regime must accept the bulk of responsibility for acceptable inflation performance.

The first thing to note is that under the institutional set-up which operated well until the mid-1970s, monetary policy was highly constrained in what it could achieve. The fixed exchange rate was one key factor. Figure 6 shows growth in M3, which was for many years the principal monetary aggregate,^[15] and the change in the central bank's holdings of foreign reserves, as a proportion of M3. While a number of other factors go into monetary growth as well – the expansion of the domestic assets of the central bank, bank lending, etc. – the picture in the 1950s in particular is quite consistent with the idea that shocks originating from the balance of payments and impacting on the economy through the exchange rate mechanism were quite large, and there is an obvious correlation between growth of money and the changes in foreign reserves.

The central bank had open to it a range of tools – reserve requirements, suasion over bank lending and so on – which were used to try to ameliorate the effects of these foreign shocks on the domestic financial system, and of course to pursue other objectives as well. But there was inevitably a certain passivity in monetary management, which reflected quite clearly the textbook propositions about monetary dependence under fixed exchange rates. This was less the case in the 1960/61 recession, which has gone down in history as a domestically generated credit crunch, but even there the balance of payments mechanism worked strongly to produce the monetary contraction, as noted in section 2.^[16]

It is commonplace to ascribe the rise in the average rate of inflation in the 1970s to an increase in the growth of money. The facts are certainly consistent with this for Australia, and the analysis of the episode in question in section 2 emphasised money growth as an important channel for importing foreign inflation into Australia in this episode. More generally, the role of money played an integral role in a generation of theoretical and empirical models, particularly those developed at the Reserve Bank.^[17] A sharp fall in monetary growth in 1974 also reflects the role of monetary policy, broadly defined as including capital controls at that time, in conjunction with other forces, in slowing the inflationary boom.

The close connection between foreign forces and monetary growth lessened after this, even though capital mobility continued to be a short-term problem while ever the exchange rate was fixed. The growth of large budget deficits in the mid-1970s, and more rapid growth in bank lending, became much more important as ‘determinants’ of money growth.^[18]

It was still difficult for domestic monetary policy to contain monetary growth, however, and it became increasingly clear in this period that the institutional framework for monetary policy was inadequate. Foreign capital flows proved to be an irritant to any attempts at controlling domestic liquidity. Direct controls over banks were ineffective against the very large non-bank intermediary sector, and they were not always effective on banks themselves, since banks could not always control the growth of their lending under the overdraft system with regulated interest rates. Interest rates on government securities were still administered, and government debt management issues impinged on monetary policy. Hence, for anyone who views monetary growth as crucial for inflation, it is unsurprising that policy in Australia in the latter 1970s was not fully successful in restoring low inflation. This can also be seen by looking at interest rates, which were low or negative in real terms for most of the 1970s (see Figure 7).

Against that background, it was a pragmatic desire for more effective instruments, as much as philosophical support for market-determined prices and resource allocation, which helped the deregulatory push gather steam at the end of the 1970s. Monetary policy played an important role in reducing inflation in the 1982–83, but it was not until the end of 1983, when the link between government debt management and monetary policy had been severed by the introduction of the tender system for debt sales, and the exchange rate allowed to float, that the set of instruments was available for monetary policy to be fully effective.

Ironically, measures which were seen as enhancing monetary control were also associated with other regulatory and institutional changes which led to pronounced shifts in the relationships between most of the measures of money and credit and nominal income or prices. Measuring the stance of monetary policy simply by looking at the growth of the aggregates became much more difficult in the second half of the 1980s.

A full treatment of the indicator properties of monetary and credit aggregates is given in Blundell-Wignall et al. (1992). For the present paper, two simple illustrations of the problem suffice. The first is in Figure 8, which shows growth of M3 against prices. The data are smoothed by taking a centred 11-quarter moving average in both cases, to remove short-term fluctuations. It is clear that money growth was quite correlated with the pick-up in inflation in the late 1960s and early 1970s. But it gave a seriously misleading signal for much of the 1980s. Indeed, the graph looks like a mirror image in that period.

The second illustration comes from taking averages of growth for the whole range of aggregates over a longish period (Table 2). This does not seem to rescue the simple use of monetary or credit growth as a guide to inflation.

The table is highly suggestive that over fairly long periods in the 1980s, growth of the aggregates have not been good guides to average rates of inflation. That is not to say that the aggregates are useless altogether. In fact, in most instances they still have a pronounced cyclical pattern which is correlated with the cyclical pattern of output or nominal income. It is this property which shows up in the results of Blundell-Wignall et al. (1992). That is, the turning points in the aggregates may provide useful information. In contrast, it was the indicator value of the absolute rate of growth, as a simple guide to the likely underlying rate of inflation, which was in doubt in the 1980s.

Since 1990, all the aggregates have slowed sharply, and associated with this has been a contraction in output and a fall in inflation. No doubt future empirical researchers will find a much improved correlation between the aggregates and inflation when this period is included in the data set. Whether the actual rates of growth are now more significant than they have been for some years remains a matter for conjecture. (If they are, they suggest strongly that inflation will remain low.)

If the aggregates are not always helpful, what indicators can be looked at to gauge the stance of monetary policy? At a statistical level, that is not a question which can be dealt with at length here. Blundell-Wignall et al. (1992) examines the question of leading indicators of inflation in a rigorous fashion.

At a practical level, in the search for a more immediate indicator, policy makers themselves have tended to focus more on financial prices. The level of nominal short-term interest rates is the central bank's principal instrument, and therefore gets considerable attention. Due consideration must be given to the crucial distinction between nominal and real rates, with the problem being how to measure the latter, and how to judge how ‘appropriate’ they are given the stage of the business cycle, since no theory would suggest that real short rates should not vary over the course of the cycle. The measurement problem is possibly lessened by looking at the slope of the yield curve, on the assumption that long-term yields embody the financial markets' best guess of future inflation. The yield curve measure also contains something of the Wicksellian idea of the difference between the financial rate of interest and the natural rate of return on real capital as being the driving force for inflation or deflation. If long-term yields are driven by market forces to be in line with the natural rate, this offers some guide for the central bank in where to set its instrument on average over the business cycle, depending on whether inflation is too high, too low or about right. There is also some evidence that the slope of the yield curve does have some predictive power for real economic activity and inflation; see Blundell-Wignall et al. (1992), Lowe (1992).

On either measure, some proxy for real short-term rates or the slope of the yield curve, the overall stance of monetary policy over the past half dozen or so years has generally been a disinflationary one, with some periods of more accommodative stance (such as from mid-1987 to early 1988) the exception rather than the rule.

Much has been made of that particular episode, with the implication that the easing in monetary policy in the first half of 1987 was the dominant factor responsible for the very strong growth in the economy in the subsequent two years. The circumstances of the time were that the exchange rate – the principal proximate threat to inflation – had stopped falling and begun to rise; the real economy was flat; wages appeared to be remarkably well controlled – they were falling in real terms; indicated rates of inflation were peaking, and all prospects appeared to be for a slowing. Some easing in financial conditions was warranted: they were exceptionally tight, and had reached that degree of tightness essentially in support of the exchange rate, which no longer needed that support.

It cannot be true, moreover, that the strong growth of 1988 and 1989 was all attributable to the 1987 easing. With the state of the world economy and the terms of trade, it would have been quite remarkable indeed if growth in Australia had not picked up substantially.

Ultimately, perhaps one can get away from this problem of indicators when evaluating policy ex post, and simply look at the underlying rate of inflation itself, since it can still be accepted that monetary policy will influence that most strongly in the long run. The evidence here is that inflation has fallen over the 1980s, with some setbacks along the way. If this suggests anything about the stance of policy, it must be that it was in a disinflationary direction on average, though one might, of course, take issue with the rate of progress. The slow pace of progress in reducing inflation below the 5 per cent achieved at the end of 1984, which is apparent in comparisons with the G7 average, reflects a number of factors. High on the list would be the nature of the shocks hitting the economy (especially the terms of trade fall and associated fall in the exchange rate in 1985–86, at a time when other countries benefited substantially from falls in oil prices), and the time taken in subsequent years to wind this inflationary impetus out of the system.

A counter-argument which is heard increasingly is that though consumer price inflation fell over the course of the 1980s, asset price inflation was grossly excessive, especially in the latter part of the decade. While there were large rises in asset prices, the suggestion that movements in asset prices be a criterion for evaluating the effectiveness of monetary policy raises a number of questions. For example, it is often forgotten that the first asset prices to rise strongly were share prices. A strong rise in share prices began in the mid-1980s during the initial phase of recovery from the 1982–83 recession, associated with a restoration of business profitability. This preceded the property price boom in Australia, which only really got going in full strength after the slump in share prices in 1987. Was it monetary policy's job to target share prices, or other asset prices for that matter?

Surely the answer is no. Apart from the fact that asset prices may change due to things unrelated to monetary policy concerns, monetary policy can only hope to tie down one relative price: the price of money in terms of something else, either another currency (fixed exchange rate), a commodity (e.g. gold), a basket of commodities or goods and services (e.g. the CPI), or some other asset class. It is asking too much for it to tie down two relative prices – such as the price of money in terms of goods and the price of money in terms of, say, office buildings. Even if there is a dependable relationship between goods prices and asset prices in the long run, the relationship is obviously quite fluid over the period – a few years – relevant for decisions about monetary policy. This suggests that having explicit objectives for asset prices in their own right may be counter-productive.

That is not to say that asset prices should be ignored. On the contrary, they may provide information – continued high rates of asset price inflation, for example, may well have implications for general prices, through wealth effects operating on spending and so on (though the results in Blundell-Wignall et al. (1992) suggest that these effects may be hard to find empirically). This possibility was the basis for the concern with asset price rises in the late 1980s. Changes in asset prices may be an important part of the monetary transmission mechanism – in a general sense, they are at the heart of it. But the major criteria for evaluating the medium-term effectiveness of policy should be goods and services prices.

4. Costs of Disinflation

A conventional treatment of the disinflation process might begin in terms of a fairly general wage-price block, such as given in [1] and [2]:

Where is productivity growth; U is a capacity variable – unemployment, or the deviation of unemployment from the natural rate or NAIRU, or GDP less capacity; p^m is import prices; and (L) denotes a general lag structure. If the expected rate of inflation, is proxied by lags of prices, the two equations are close to a VAR system in prices and wages, with some added variables to take account of the openness of the economy and the short-run non-neutralities of policy.

This general framework allows consideration of a number of potential issues of interest for policy makers: the slope of the short-term trade-off (i.e. the short-term responsiveness of wages and prices to excess capacity in the labour and/or product markets); the location in wage-unemployment space of the long-run Phillips curve or the value of the NAIRU; the speed with which the short-term trade-off can move upwards or downwards – i.e. the speed with which inflationary expectations can adjust, and the degree of inertia present in price and wage setting; and the extent to which the NAIRU shifts during the period in which disinflationary policies run their course as a result of the policies themselves – i.e. the degree of hysteresis.

This framework accords no explicit role to the stock of money or to any fiscal policy variable. But their role is implicit in the real variables, insofar as policies are non-neutral in the short run.^[19] In the case of monetary policy, an announced, perfectly credible policy of disinflation would in theory immediately reduce and therefore actual wage and price inflation, with no associated fall in output or rise in unemployment. In practice, there seem to be few examples of costless disinflation,^[20] and the issue of interest in the present paper is the size of those costs.

There are several ways one could proceed. One is to estimate [1] and [2], and focus on the key parameters: β and ϕ, which govern the responsiveness of wages and prices to fluctuations in output and employment, and α and λ, which govern inertia in wage and price setting. Estimates of these coefficients could then form the basis of computations of the cost of reducing inflation, and the period over which that cost must be borne. A more sophisticated alternative would embed this exercise within a full-system model, and conduct a simulation. A final option, which is quite crude but which perhaps has the advantage of being more transparent, is to look at the relevant episodes in the time series, and to try to assess the extent of lost output and higher unemployment directly.

It seems worthwhile to pursue all these approaches to some extent. It is worth noting at the outset, however, that these sorts of calculations will reflect the average experience over some particular period in history, a history which is dominated, moreover, by fluctuations in prices and output which took place over the course of business cycles. Whether the statistical correlations which are thrown up by such episodes provide the means of answering an abstract question about the costs of moving from one steady-state rate of inflation to another is open to question. But since this history is the only set of facts available, there is little choice but to proceed in this manner. With that caveat clearly in mind, the next section looks at summary statistics from the major episodes. The following sections look at some issues for wage and price-setting in particular.

(a) Key Episodes of Falling Inflation: Implied ‘Sacrifices’

Table 3 summarises the major statistics of the identifiable episodes of falling inflation. It includes one episode not discussed at length in section 1 – a fall in inflation in 1956–58 – and one disinflation episode which is clearly not yet completed – namely the present period. In some instances it makes more sense to treat episodes together, as in the mid-1970s for example, when there were two noticeable falls in GDP a year apart, but when inflation fell (after adjusting for changes to health care arrangements) more or less monotonically over several years. Arguably, the contraction of output and further falls in inflation in 1977–78 should also be lumped in with the earlier two periods, and the whole period from 1974 to 1978 treated as one long effort at disinflation. That approach is adopted here.

Falls in the rate of inflation have varied anywhere from a couple of percentage points and quite temporary, to 25 percentage points, most of which was more or less permanent. In most cases, the fall in output was short-lived: real GDP regained its peak within a couple of years, and within a couple of quarters of the resumption of growth. The peak in employment was regained within a year in most cases. The costs of disinflation were, of course, considerably higher than these calculations suggest, since real GDP often remained below capacity for somewhat longer, and the rate of unemployment was slow to fall back to pre-recession levels.

The table computes some elementary ‘sacrifice ratios’.^[21] One issue is how to capture the ‘permanent’ component of the fall in inflation. Measuring the fall in inflation as the difference between the local maximum and the subsequent local minimum in the four-quarter-ended rise in consumer prices understates the sacrifice ratio, since ‘underlying’ inflation will rarely have fallen by that much, there being presumably some cyclical fluctuations in profit margins. Nor is it straightforward to get a good idea of underlying inflation before and after the disinflation, since the various episodes tended to run together. The convention adopted here is to take the fall in a three-year-average inflation rate as a reasonable indicator of the shift in underlying inflation.

It need hardly be emphasised that these calculations are quite crude. That said, they suggest that the GDP cost associated with a lasting one percentage point fall in inflation has been between 0.5 and 2.0 per cent of a year's GDP. This is a fairly wide band, but perhaps two (not unrelated) patterns emerge from the episodes. One is that in cases where foreign shocks played a very large role in pushing up inflation – the early 1950s and the mid-1970s – the fall in inflation which follows is less costly than in other episodes. The second pattern is that reducing inflation from low rates – as in the mid-1950s and the early 1960s – looks to be more costly than reducing it from higher rates. This is apparent in Figure 9.

In mid-1992, computation of the cost of the early 1990s disinflation cannot be completed. Even if inflation falls no further, the full cost of the reduction to date will not be clear until capacity output is regained and the post recession path of unemployment can be observed (for several years). But even at this stage, it is reasonably clear that the costs rank alongside those of the early 1980s and the early 1960s in size.

A pattern which emerges from consideration of the unemployment statistics is that the apparent unemployment cost of reducing inflation has risen over time. Figure 10 shows some pertinent statistics. The rate of unemployment in the 1960s fluctuated between 1½ and 2½ per cent. In the mid-1970s, it rose sharply, and finished the decade at around 6 per cent. The 1980s and early 1990s have been characterised by a further sharp rise to about 10 per cent in the early 1980s recession, followed by a gradual decline back to about 6 per cent by the end of the decade. With the onset of another recession in 1990, unemployment has risen again to about the same rate as in 1983. Peaks in unemployment, and the corresponding cyclical low points, have grown successively higher through each cycle. The duration of the period of high unemployment has also grown larger over time. The crude unemployment sacrifice ratios reflect both these facts, rising strongly through successive episodes.

This pattern might be due, in part, to changes on the supply-side of the labour market. There are reasons to think so in the 1974–78 episode, and a sacrifice ratio cannot be sensibly calculated for this case because unemployment remained permanently much higher than in the 1960s. This should have been less of a problem in the 1980s, since real wages showed some downward flexibility. Even so, the sub-6 per cent rate of unemployment of 1980 was not achieved again until late 1989. Both the extent of the rise in unemployment in the 1980s and the duration of unemployment above anything like its previous level were unprecedented, apart from the mid-1970s when, as noted above, there were big supply shocks.

One obvious question is why this should be. It is conventional to see the answer to this as lying in wage inflexibility. There is much in this for the 1970s, as argued in section 2 of this paper. But, as noted above, it is far from obvious that this was the case in the 1980s – the period of falling real wages over several years in the mid-1980s was remarkable. This provides an obvious entry into a brief discussion of wage setting.

(b) Wages: The Phillips Curve, the NAIRU and the Accord

There was a long and fairly vigorous debate on the Phillips curve in Australia in the 1970s, focussing on the question of whether a long-run trade-off existed (the general conclusion was that it did not) and on technical issues relating to uncovering that conclusion and the related issue of estimating the NAIRU. Hagger (1978) reviews these issues at length. Even as this debate was being waged, the short-run Phillips curve was apparently in the process of shifting, as the events of the mid-1970s pushed unemployment/inflation combinations well outside the range of outcomes which had been known in the post-War period. Parkin (1976) noted this almost as a post-script in his final contribution to this section of the Australian debate.

Figure 11 shows a scatter plot of the rate of wage inflation and the rate of unemployment on a quarterly basis since 1960. The black squares are the observations between 1960 and 1970. They suggest something of a negative slope. Inflation was low on average and not particularly variable in this period, as discussed in section 2 of this paper. It is perhaps not unreasonable, therefore, to draw a short-run Phillips curve on the assumption that inflationary expectations were roughly constant for most of the period. Visually, the plots seem to suggest a NAIRU of not less than 2 per cent, which is basically consistent with many of the early econometric estimates.^[22]

In the late 1960s, the observations tended to cluster at the top end of this group. Through the first half of the 1970s, the observations (joined by a line) describe a clockwise loop – rising more or less along a vertical line at 2 per cent unemployment until 1974, then falling at a somewhat higher rate of unemployment after early 1975. For the bulk of the rest of the period, the plots suggest something of a downward sloping trade-off again, with the location shifted to the right. The exception was the period 1982–85 (linked by a line), when there was another wages push, imposition of the wages freeze, and then the re-entry from the freeze period in the early days of the Accord.

An obvious candidate for a hypothesis which would fit this picture is that the expected rate of inflation took a big rise in the early 1970s, then fell, but remained higher (and stable) throughout the rest of the sample than in the 1960s, while the NAIRU shifted during the course of the mid to late 1970s, to something like 5 or 6 per cent. A major ‘loop’ occurred in the boom and bust cycle of the 1980–84 period, but observations either side of that period seem broadly consistent with some negative short-run trade-off. The last three observations – for the June, September and December quarters of 1991 – are clearly on their own at the bottom right.

This hypothesis seems broadly consistent with some other pieces of information. Figure 12 shows a measure of inflationary expectations, calculated by the Melbourne Institute of Applied Economic and Social Research. Both the mean expectation and the median are shown. Unfortunately, these series have no history for the 1960s. But generally speaking, the available data do seem consistent with a rise in inflationary expectations in the mid-1970s as actual inflation accelerated, then a fall back but to still fairly high levels in absolute terms. Expectations on this measure were remarkably stubborn through the 1980s, and have only recently fallen noticeably. This suggests that the idea of a short-term wage inflation/ unemployment trade-off at a roughly constant rate of expected inflation for most of the 1980s – as embodied in Figure 11 – is defensible. It also suggests the location of the short-run curve is likely to be shifting downwards at present.

On shifts in the NAIRU from the supply side, there is the big rise in the real wage itself, which persisted for most of the second half of the 1970s. Other supply influences are perhaps a matter for detailed micro-economic study, beyond the scope of this paper. Figure 13 shows one piece of circumstantial evidence that the work/unemployment choice may have shifted in favour of more unemployment: the ratio of unemployment benefits to after-tax male average weekly earnings rose sharply in the mid-1970s, nearly doubling. It fluctuated somewhat in the 1980s, but remained well above its level in the 1960s. One could not suggest that all the rise in unemployment in the mid-1970s was voluntary, but there is a case for arguing that the NAIRU rose noticeably because of supply factors.

Several studies of the wages/unemployment equation in Australia are also consistent with this general picture. Gregory (1986), to pick just one, found that the unemployment rate was statistically significant in a wage equation only in the presence of a time trend; and that the coefficient on the price expectations terms was far from unity, in contrast to the writers of the early and mid-1970s. His sample period was 1966 to 1982, a period in which the NAIRU is likely to have shifted. A reproduction of Gregory's result is shown in Table 4, in the first two lines. The third and fourth lines of the table extend Gregory's sample period to 1991, retaining the simple specification. Here the unemployment rate is significant, and the trend term is no longer significant. That is, sample periods which are dominated by drawings from the 1970s do not show any influence for the level of unemployment on its own (a specification in which the NAIRU is implicitly constant), and seem to imply that the NAIRU was rising. A sample period which is less dominated by the 1970s, and particularly in which the 1980s looms large, shows a much bigger effect for the level of the unemployment rate and, of course, a much higher implied level of the NAIRU.

A specification in which the unemployment term is the difference between the current unemployment rate and a twelve quarter moving average of past rates (as a crude proxy for a natural rate displaying hysteresis) performs well in the 1966–82 period, but poorly for 1966–91. A specification in which the unemployment term is defined as the deviation of unemployment from 6 per cent performs poorly in the 1966–82 period, but much better over the period which includes the 1980s.

This is, of course, very simple econometrics. But all these results appear to be consistent with the sorts of shifts in the short-run trade-off suggested above.

Other work on the wage equation has employed a different model of the labour market, inspired by the insider/outsider distinction developed in the context of European labour markets by Lindbeck and Snower (1988), Blanchard and Summers (1986), Layard and Bean (1989). In such models, wages are not simply assigned the function of clearing the aggregate labour market. The aggregate rate of unemployment – the pressure exerted by ‘outsiders’ – is only one important factor. Of key importance is the behaviour of ‘insiders’, who bargain with employers over nominal wages with objectives for expected employment of their own group. The concept of an aggregate NAIRU may not even exist in this world.

There is some support for this hypothesis in the Australian literature with measures of excess demand for labour ‘within the firm’, such as overtime hours worked, sometimes found to be as important as aggregate variables like unemployment or job vacancies. Gregory and Smith (1983) found that the use of overtime hours worked allowed an apparently stable wage equation to be estimated. Dawkins and Wooden (1985) used a different measure of ‘internal’ labour market pressure, but found a similar conclusion. A careful study by Watts and Mitchell (1990) using the deviation of real GDP from capacity as a proxy for internal pressure found support for the insider/outsider model, but not for the conventional Phillips curve.

It is probably sensible to regard these hypotheses not as polar opposites, but as complements. Alogoskoufis and Manning (1988) found that the differences between European, Japanese and US insider/outsider dynamics were not sufficient to explain the differences in unemployment outcomes between the various countries after the mid-1970s. They tentatively concluded that persistence of macro-economic shocks – both demand and supply – were more likely to be responsible for the persistence of high unemployment in Europe. Groenewold and Taylor (1992) follow Alogoskoufis and Manning's approach with Australian data and find some evidence of insider effects, but also that this is not sufficient in itself to explain the Australian data. Simes and Richardson (1987) favour an eclectic approach, with a largely conventional expectations-augmented Phillips curve model, supplemented by overtime hours worked as an ‘insider’ pressure variable. The best conclusion seems to be that there is some support for the idea of insider-outsider distinctions, but that it is far from clear that aggregate labour market conditions are irrelevant.

One reason this may be true is the existence of incomes policies in Australia in the 1980s. In the context of insider-outsider distinctions, the Accord can be seen as a way of internalising labour market externalities – with the power of insiders to affect wages conditioned by the social compact which formed the Accord, so as to permit lower nominal wage growth and higher aggregate employment for a given rate of inflation than would be the case with the insiders left to bargain freely. In this view, the Accord did help to alter the shape of the short-term unemployment-inflation trade-off. The comprehensive study by Watts and Mitchell (1990) concludes quite clearly that incomes policy did reduce nominal wages growth.

The contrary view is that the Accord was a veil, and reproduced what a properly-functioning market would otherwise have delivered. The clearest statement of this position is in Pissarides (1989).

The general question can be put in a nutshell in terms of Figure 11: do the observations when the Accord was in force fall in a way which marks them out from the general short-term trade-off between wages and unemployment? The Accord came into force about half-way through the 1981–85 loop, when the trade-off looked as though it had shifted further to the right after the 1982 wage explosion. From early 1985, outcomes looked more in line with the overall pattern of the 1975–90 period. The Accord might have helped produce this apparent shift back to the left, but it still might have happened through natural adjustments to the preceding shock.

To look at the latter period, Figure 14 reproduces Figure 11, with observations from March 1986 to March 1991 shown as black triangles. It is suggestive of the idea that the short-term trade-off was affected by the Accord: unemployment ranged over nearly 3 percentage points in this period, including to rates which were reasonably low by the standards of the 1980s in Australia and other countries. Wages growth remained at around 7 per cent throughout. In fact, one might offer the suggestion, only partly tongue-in-cheek, that there was a horizontal trade-off between wage inflation and unemployment in this period, at a ‘natural’ rate of wage inflation of 7 per cent. This is essentially the assessment of Carmichael (1990), who argues that inflation was set by wages policy in the 1980s, with monetary policy in a subordinate role.

In this perspective, the Accord did operate to alter the slope of the short-term trade-off, as might be expected of an incomes policy. But the focus was more on enhancing short-term growth and employment for a given rate of inflation, than on lowering the transitional costs to growth and employment of reducing inflation. That is, the Accord flattened the trade-off, but did not shift it down.

A strong version of this hypothesis would have the corollary that the Accord actually increased the cost of reducing inflation below 7 per cent (or whatever rate of price inflation goes with 7 per cent nominal wage growth). But this does not seem to be the case, since over the past year or so nominal wage growth has broken out of the 7 per cent range, again apparently tracing out some sort of short-term trade-off.

So we are left with the role of the Accord being difficult to determine from much of the observed data. What can be said most clearly about the Accord, in my judgement, is that it did operate to prevent wage growth accelerating sharply in 1988 and 1989. Wages growth picked up only very slightly in this episode, in contrast to the other periods of similar strength in the demand for labour in 1981–82. Even though private investment – a key driving force in Pissarides' model-picked up sharply, and in fact as a ratio to GDP reached a post-War record high, nominal wages growth barely kept pace with prices. In the context of Figure 14, it is highly likely that the observations towards the left-hand edge of the set of Accord outcomes might otherwise have been displaced upwards had the Accord not been in force. Other policies may also have helped. Wage-tax trade-offs, in particular, may have reduced nominal wage growth. There are few definitive tests of this, and the evidence from other countries provides little encouragement for the idea that tax changes affect wage outcomes in this way, but preliminary results in Blundell-Wignall and Stevens (1992) do suggest some support for it in Australia.

(c) Price-Setting Behaviour

Unfortunately, in contrast to the detailed study of labour market functioning and wage setting, there is not much in the Australian macro-economic literature about price setting. This section cannot remedy that deficiency, but a few important stylised facts can be illustrated.

From Figure 4, there is, as expected, a fairly clear association between wage and price growth over long periods. When looking at the turning points, there are cases of wages leading prices, and others of prices leading wages. A tendency for wages to grow, on average, faster than prices, is also quite marked, with wage growth dipping below price growth in periods of recession – the early 1950s, 1957, the early 1960s, 1983. What is striking about the graph is the way in which the two series apparently became de-coupled in the period between 1982 and 1990. The wage freeze, and then the re-entry after that, produced sharp fluctuations in real wages. From 1984 to 1990, nominal wage growth was around 5 to 7 per cent, similar to the mid 1960s. Yet consumer price inflation was well above the mid-1960s average of around 3–4 per cent.

It is also apparent that consumer prices have more inertia than do nominal wages. On the surface, at least, this suggests that pricing behaviour over periods as long as a year or two (or perhaps longer) cannot be assumed to be just a markup over wage costs. In fact, estimating equation [2], it turns out that wages can often be excluded with little loss of performance.

Rather than try to fully capture the complexities of wage-price linkages, the following discussion focuses on a very simple equation which follows from fairly standard assumptions:

where is the price of domestically-produced goods, and the price of tradeables, proxied by imports. ω is the share of tradeables in consumption. Prices for domestically-produced goods are given by:

where E( ) denotes expectations, and U_t is a capacity variable.

If expectations are conditional on all past prices:

and

This equation, which can also be thought of as a reduced form of [1] and [2], is along the lines of that used by Cozier and Wilkinson (1991) in their discussion of the costs of disinflation in Canada, the main difference being the inclusion of the open-economy aspects via import prices. It is also similar to that used in Blundell-Wignall et al. (1992) in their analysis of inflation indicators. Interpretation of the results as providing information about the costs of disinflation depends on the assumption that the output gap is the independent variable – i.e. that policies are non-neutral, affecting output, and that prices are sticky. This assumption has been maintained throughout, but it is worth noting that some might instead interpret equation [3] as a Lucas aggregate supply function, in which the deviation of prices from their expected level leads to temporary fluctuations in the supply of output.

Results of estimating equations of this form are reported in Table 5. The quarterly rate of inflation is regressed on past prices, import prices, the output gap and a wage shock variable. The adjustment to the co-variance matrix suggested by White (1980) is used to derive the standard errors. Four lags of the right-hand side variables are used in each case and, in the case of import prices, the contemporaneous term is included as well. The output gap is the deviation of GDP from a trend computed using a Hodrick-Prescott (1980) filter. This filter allows the growth rate of trend output to vary over time. The intuitive appeal of this is that it potentially copes better with supply shocks which affect growth of potential output than would a simple linear trend.^[23] The wage shock variable is the residual from a wage equation where wages are explained by past prices and the output gap. Thus it is the component of wages growth which cannot be explained in terms of a standard Phillips curve sort of wage model, and which might be put down to supply shocks.^[24]

Generally speaking, most of the results are intuitively plausible. Explanatory power is reasonable for equations specified in rates of change. There is no apparent problem with serial correlation. The share of imported consumption goods in total private consumption is about 5 per cent, while the ‘wholly or predominantly imported’ components of the CPI make up about 10 per cent. As a benchmark, one might therefore expect the coefficient and import prices to be something like 0.05 to 0.10. This is usually the case. The lags of prices generally sum to about 0.75–0.8. Ideally, one would like this sum to be about 0.9, and the import price and lagged total price coefficients to sum in total to one. This would be consistent with a model in which inflation was driven primarily by inertia and foreign price movements, and was pushed off this path by output gaps or supply-induced wage shocks. The ideal model would not have a significant constant term, since a constant suggests that there is some unexplained factor driving prices beyond those mentioned above, and it would also suggest that the output cost of reducing inflation below some critical level would be extremely large.

As it is, a significant constant term is found in the regressions including the 1960s, usually with a value equivalent to an annual inflation rate of between 1 and 2 per cent. It may be that this reflects some non-linearity in the short-run trade-off, some minimum rate of inflation which is for practical purposes irreducible. It could be due to the effects of quality change on price series, which is usually thought to produce an upward bias in measured inflation rates. Alternatively, it might reflect part of the expectations mechanism, particularly in the 1960s, when expectations of inflation were low and constant. That is, the constant term may be doing some of the work of the lagged price terms in picking up expectations.

The wage shock term plays a role in the estimations including the 1960s and early 1970s. In the 1976–1991 estimations, however, it is insignificant. This appears consistent with the apparent weakening of the relationship between wages and prices in the 1980s noted earlier.

The coefficients on lags of the output gap add to something like 0.1. The hypothesis that they are jointly zero is rejected at the 13 per cent level of significance for the full period estimation, and the 5 per cent level in the 1976–91 period. The hypothesis that they add to zero is rejected quite strongly in every case.

Tests for asymmetry – i.e. whether below-trend GDP has more downward effect on prices than above trend GDP has upward effect, and whether periods in which GDP is actually falling have an extra downward effect on prices – suggested no particular pattern (these results are not shown here).

The equations in Table 5 suggest that an output gap of 1 per cent for one year has been associated with a lasting reduction in inflation of something less than half a percentage point (the temporary fall in inflation may be bigger). In other words, the sacrifice ratio has been about 2½–3 based on most of the equations here. This particular statistic is sensitive to the estimated lag structure, the nature of which varies over different samples. Some subsample estimations give a sacrifice ratio of less than 2; others as high as 4. So there is a reasonably large margin of uncertainty about these estimates.

There are, of course, all sorts of other caveats which need to be applied to this very simplified analysis. One is that the measurement of capacity gaps is imperfect. Another is that terms of trade changes in the open economy mean that consideration of only consumption prices is incomplete. The analysis could also be criticised for not examining questions of monetary policy changes directly. For all these and other reasons, the results are by no means definitive.

Having said that, it is worth comparing them with the results from other techniques. Table 6 summarises estimates of the cumulative output losses involved in reducing inflation by one percentage point permanently based on the above equations with the apparent sacrifice ratios from Table 3, and those quoted in EPAC (1991), based on the Murphy model, one of the best known macro-econometric models of the Australian economy. Murphy conducted a simulation analysis, where a monetary policy tightening was announced and implemented, so as to bring about a 3 percentage point fall in inflation in the long run. The sacrifice ratio implied in the base-case scenario was 2.3 for GDP, and 0.9 for unemployment. In a ‘low credibility’ alternative (in which financial markets did not immediately fully adjust expectations in line with the announced price objective), the sacrifice ratios were only slightly higher (2.5 and 1.0 respectively). The real difference to sacrifice ratios came in a scenario in which the lags in the prices/wage block were shortened substantially; in this instance, the costs of disinflation were, of course, much lower than in the baseline case.

To the extent that there is any consensus emerging from all of this, it seems to be that a reduction of 1 percentage point in inflation involves lost output the equivalent of between 1½ and 3 per cent of a year's GDP. These numbers are somewhat higher than the estimates for Australia obtained by Andersen (1992), which may well serve to underscore the sensitivity of these sorts of calculations to the way in which potential GDP is estimated, and the time periods involved.

The unemployment sacrifice ratio has an even wider range of estimate. The equations in Table 4 suggest that wage growth falls by 0.2 per cent per quarter for every 1 percentage point rise in unemployment, given price expectations. On the assumption that this is reflected fully in lower price growth within a year, and that price expectations of wage bargainers adjust fully (or there is little inertia), this would imply a sacrifice ratio of about 1, which is close to EPAC's result. With only partial adjustment of price expectations (as in equation 4.20), the sacrifice ratio is over 2, and not far from EPAC's ‘low credibility’ outcome (though for slightly different reasons).

Both these and the EPAC estimates of the unemployment sacrifice contrast with the estimate based on what actually happened in the 1980s, where the apparent cost of reducing inflation was much higher. This might be taken as evidence of expectations being slow to adjust, and so a lack of credibility in anti-inflation policy. It would have to be conceded that, for one reason or another, inflation expectations have until recently been remarkably stubborn. The question is whether this is the whole story, or whether the time path of unemployment is such that other factors – to do with the nature of the unemployment experience in severe recessions – have some importance. If the latter is part of the answer, it suggests that the costs of a concerted disinflation effort, while perhaps not permanent, can nonetheless persist for some time.

5. Conclusion

Australia saw a big rise in inflation in the mid-1970s, along with other countries. In fact, this paper has emphasised the importance of foreign shocks for Australia's inflation rate. What was remarkable about the mid-1970s was not the magnitude of the rise in inflation – that was not unprecedented at all, as a glance at the 1950s demonstrates – but the fact that high inflation persisted, only being finally reduced to quite low levels by the early 1990s. Other countries had difficulty in combating inflation too, but Australia's performance deteriorated relative to other countries in this period – it went from being a relatively low inflation country in the 1960s, to a relatively high inflation one in the 1970s and 1980s.

The period of maximum divergence from the G7 average was between 1985 and 1988. Yet the 1980s was essentially a decade of disinflation. Viewed from the vantage point of 1992, the trend in inflation is clearly down over the period since 1982. Looking over the long run from 1950 to the early 1990s, it is possible in fact to see the outlines of a few broad trends. The 1950s was mostly a period of disinflation after the imbalances of the late 1940s and the Korean War boom. The mid-1960s, without large foreign shocks, was a golden age for the Australian economy, with strong output growth and steady, low inflation. The period from around 1968 to 1975 was essentially one of increasing inflationary pressure, and rising inflationary expectations. Between 1975 and 1982, the extreme rates of inflation were wound back, but the core rate of inflation proved more difficult to shift below 8–9 per cent. Several bouts of tight monetary policy – briefly in 1974, more aggressively in 1982, and more aggressively still and more consistently later in the decade, with an inverse yield curve most of the time from 1985 to 1990 – were required to make solid progress against the inertia of the inflationary process. Incomes policy played some role in this process as well, especially at a crucial stage in 1988 and 1989.

This was not really all that different from the general story in other countries, except perhaps that they did more of the job earlier. They were also helped by the commodity price slump in 1985–86, which proved to be a substantial setback to Australia. More than any other single event, the fall in the exchange rate in mid decade made comparisons of Australian inflation with that of other countries look poor. The fact that Australian inflation might be above that of the rest of the world for two to three years as part of a business cycle disturbance should be unremarkable if inflation started at the world rate and then returned to it afterward (and, for the sake of symmetry perhaps, went below it for a period). But on this occasion, the fact that Australia was already lagging a little in the disinflation process – as it had lagged in the upswing – as well as undergoing some structural realignment in the real exchange rate, meant that the comparison with the G7 painted a very unflattering picture indeed for several years.

The behaviour of the exchange rate since the mid-1970s – a pronounced downward trend – in fact raises an issue that this paper has not touched on: the short-term conflict between an anti-inflation strategy and balance of payments objectives. It makes sense to think of this as an open economy extension to the simple Phillips curve world. There is no long-term trade-off between real variables (output, current account) and inflation. But there is some short-term trade-off. There is always some temptation to exploit the short-term trade-off for more real quantities by inflating (or, the same thing, devaluing the currency). By the same token, there is usually some short-term cost to disinflation – lower output, a higher real exchange rate and therefore stress on the traded sector. This paper has not examined the question of the distribution of the output costs of disinflation, which might fall disproportionately on the trade sector, but it is an on-going issue in Australia.

On the costs of disinflation in general, the empirical evidence available suggests that a percentage point reduction in inflation may entail an initial cost of up to 2½–3 per cent of a year's GDP in lost production, with a considerable degree of uncertainty attached to that ‘guesstimate’. It is stating the obvious to say that the role of policies in affecting inflationary expectations, and through that price and wage setting directly, is of key importance in the process. The cost in terms of unemployment is more difficult to define analytically since the price-wage link appears to have been looser in the 1980s than in the past, but simple observation of the 1980s suggests that it is quite high, since upward shocks to unemployment appear to have been quite persistent in Australia (as elsewhere).

There do not seem to have been any examples of costless disinflation. The closest to this ideal was probably the initial fall in inflation in the mid-1970s from its temporarily very high rate. There was a fall in GDP and a rise in unemployment, but allowing for the fact that these must have owed something, perhaps a lot, to supply-side shocks, it could be argued that the true sacrifice was small. That disinflation was helped, of course, by the stabilisation of world inflation (though Australia did not derive full benefit from that because the exchange rate declined). It is important to note, however, that this period did not succeed in pushing inflation any lower than about 8 per cent. An important question also remains as to whether the costs of disinflation are invariant to the starting point for inflation. Looking across the set of episodes, it would appear not.

Lowering the costs of disinflation is clearly on the agenda. Here Australia will need to observe experience elsewhere carefully. The experience of the United Kingdom in the European Monetary System, and the experience of Canada and New Zealand with explicit inflation targets, will be of interest. In each case, policy makers are seeking a credibility boost in an effort to change expectations and behaviour. It is too early yet to draw any firm conclusions, but in none of these cases has recent disinflation been obviously cheap.

The fact that the question of how to run monetary policy in a low-inflation world is on the agenda at this Conference, and more widely in Australia, is a remarkable sign of the times. The contrast between the present situation – inflation of 2 per cent in the latest year, indicators of inflationary expectations decisively reduced – with that prevailing even only two years ago, when a major survey of inflation was conducted (Carmichael (1990)), is striking. It is to be hoped that all of this portends a sustained period of general price stability.

Appendix: Data Sources

Consumer Price Index	Analytical CPI series, ABS Cat. No. 6401.0, excluding mortgage interest and consumer credit charges, adjusted for Medibank/Medicare.
Consumer Price Indexes	Foreign countries. Data source for US and UK. 1900–1960 is Liesner(1985). Source for G7 data from 1960 is OECD Main Economic Indicators.
Other Price Deflators	Imports, exports, private consumption and GDP deflators from ABS Quarterly National IncomeAccounts, Cat. No. 5206.0, December Quarter 1991 issue.
Average Weekly Earnings	For 1950–1959:3, Male AWE, ABS. For 1959:3 81:3, Male Equivalent AWE, ABS. For 1981:3–1991:4 all persons AWE from ABS Cat. No. 6302.0.
Unemployment Rate	In Figure 10, for 1950–1957:3, recipients of unemployment benefits divided by number of recipients plus number employed. For 1957:4–1966:2, unemployment benefit applicants divided by applicants plus number employed. Data taken from Commonwealth Bank Statistical Bulletins. For 1966:3–1991:4, seasonally adjusted unemployment rate from ABS Labour Force Survey, Cat. No. 6203.0. For Table 3, annual data from Foster and Stewart (1991) for 1950–66; thereafter quarterly data as above.
M3, change in RBA foreign reserves	Source is Foster and Stewart (1991), and Reserve Bank Bulletins, various.
Interest Rates	90 day bank bill rate and 10-year Commonwealth Government Bond rate from Reserve Bank Bulletins, various.
Inflationary Expectations	Westpac-Melbourne Institute Survey, Westpac Melbourne Institute Centre for Business Cycle Analysis.
Benefit Ratio	Ratio of after-tax unemployment benefit to after-tax male average weekly earnings (using earnings in Table 4.17, unemployment benefits in Table 2.25, and tax rates in Table 2.23, in Foster and Stewart (1991)).
Real GDP	GDP at 1984/85 prices from Foster and Stewart (1990) annually for 1949/50 to 1959/60, and quarterly from 1959:3 from ABS Cat. No. 5206.0, December Quarter 1991 issue.
Output Gap	Real GDP is detrended using the method suggested by Hodrick and Prescott (1980). This is described in detail in Blundell-Wignall et al (1992). This procedure was applied to annual data for the log of real GDP for 1949/50–1959/60, and for quarterly data for 1959:3–1991:4. The output gap is then the difference of the log of actual GDP and detrended GDP.

Footnotes

Economic Analysis Department, Reserve Bank of Australia. Research assistance by Robert Subbaraman is gratefully acknowledged. Geoff Heenan provided assistance with data, and Luci Ellis with understanding aspects of the industrial relations system. A number of colleagues provided helpful comments on an earlier draft, particularly Adrian Blundell-Wignall, Phil Lowe and Malcolm Edey. The above, and the Reserve Bank of Australia, bear no responsibility for the views in this paper, which remain my own. [*]

To complete a series for consumer prices back to 1950, the Consumer Price Index (CPI) is used for periods prior to the commencement of quarterly national accounts data in 1959. For later periods, the consumption deflator is preferred to the CPI to avoid measurement difficulties in the latter, especially changes to the method of financing medical insurance in 1984 and the effects of the introduction of mortgage interest charges in 1986, which have affected the short-term reliability of the published CPI as an indicator of prices for current consumption goods and services. [1]

The Australian pound traded at parity to sterling until 1931, when it was devalued by 20 per cent as part of the strategy to deal with the Great Depression. The pound was devalued against third currencies with sterling in 1949, but the $A was not devalued with sterling in 1967. From December 1972, the $A was fixed against the US dollar; from September 1974 against the TWI. The crawling peg, with daily adjustments, was instituted in November 1976; the $A floated in December 1983. [2]

The important theoretical linkages include the following:

• A rise in world prices for traded goods means that the prices of imports and exports will rise directly, assuming the country is a price-taker in international markets. This will have an immediate effect on the aggregate price index (which includes such goods).
• There may also be second round effects. Traded goods will be relatively more attractive to produce, and less attractive to consume. The reverse is true for non-traded goods. This will induce excess domestic supply of traded goods, and excess demand for non-traded goods, putting upward pressure on non-traded prices. Depending on production and consumption elasticities, there may also be a change in the current account of the balance of payments. That would mean, ceteris paribus, an accumulation of foreign assets by the central bank, and a corresponding increase in its domestic liabilities. The domestic money stock increases, reinforcing the tendency for domestic prices to rise. Output may rise temporarily; prices will rise permanently.
• To the extent that high inflation in major countries is associated with excessive rates of monetary expansion, some models emphasise capital flows out of high money growth countries and towards other countries, which would result in an expansion of the money stocks of those countries. This would imply excess demand for money at the prevailing set of interest rates and prices in the small country, and would place upward pressure on prices there (perhaps temporarily expanding output as well).

[3]

The ‘Scandinavian model’ emphasises different longer-run issues. The domestic rate of inflation is the world rate of inflation, plus the difference in productivity growth between the competing and sheltered sectors. The behaviour embedded in this model is that wages in the two sectors are the same. The model has not been widely applied to Australia; one paper on the question (Mitchell (1989)), concludes that data restrictions prevent its use. [4]

The source of the data is, for Australia, Butlin (1977), and ABS Consumer Price Index (Cat. No.6401.0), for the US and the UK, Liesner (1985). [5]

See for example Meese and Rogoff (1983). For Australia, standard tests of uncovered interest parity have often tended to give unsatisfactory results in explaining short-term exchange rate movements: see Macfarlane and Tease (1989). Blundell-Wignall and Gregory (1990) find evidence of a long-run relationship between the terms of trade and the real effective exchange rate, though the speed of convergence to the long-run equilibrium is slow. [6]

Details of the detrending procedure are in the Appendix. [7]

The following table from Waterman (1972) for an index of land prices in Melbourne gives the flavour of asset price inflation in the 1950s:

Year	Index Value
1954	328
1955	388
1956	472
1957	560
1958	709
1959	928
1960	1045
Source: Waterman (1972) p 194.

[8]

Pitchford (1977) table 4 (page 377). [9]

Decisions were not always for the full CPI increase; often they were for less. In many instances, the Commission also awarded flat amounts, rather than percentage increases. In later years of the 1970s, ‘plateau’ indexation, ‘partial’ indexation, and compression of wage relativities became issues. Rimmer (1985) gives a very good account of this period. [10]

In practice it also depended on how the wage rises were delivered. Caton et al. (1974), for example, found varying results from simulations using the NIF model, depending on such factors. [11]

Australia's experience with monetary targeting is reviewed in an international context by Argy, Brennan and Stevens (1990). [12]

The Reserve Bank's Index of Commodity Prices fell (in SDR terms) by 31 per cent between January 1985 and August 1986. [13]

Calculations based on these ABS data are likely to underestimate the exchange rate effect for a couple of reasons. First, the implicit control path is one where prices for those goods would have been stable except for the depreciation. But there is good reason to believe that the prices of imports would have been falling but for the depreciation, as pointed out by Richards and Stevens (1987). Second, the ABS method appears to ignore the effects of changes in the exchange rate on prices of goods which use imported inputs, and of exportables, which should also rise in price relative to the ‘no-depreciation’ alternative. [14]

At this stage, the size of financial intermediaries other than banks was so small that broader aggregates which would be used nowadays can be ignored, in favour of M3, for which the ‘formation table’ or ‘counterparts’ type of analysis can be applied. In these early times, such an approach made perfect sense, since the exchange rate was fixed and interest rates were either regulated or strongly influenced by central bank ‘support’ for the government securities market. [15]

Note that the statistics in Figure 6 reflect the impact on reserves of an IMF borrowing, so that the size of the contraction coming from the foreign component of monetary creation is understated. [16]

See Jonson (1976), Jonson, Moses and Wymer (1976). [17]

In the 1980s, with the moves to deregulate interest rates and the exchange rate, there appears to be, as expected, little mechanical relationship between the changes in reserves per se, and M3 growth. This is in contrast to the contention sometimes heard that the fast growth in M3 in the latter 1980s was due to foreign exchange intervention by the Reserve Bank. In fact, as has been explained elsewhere, rapid growth in M3 in that period had more to do with innovation and regulatory change. [18]

There is also a potential for changes in monetary or fiscal policies to affect import prices through the exchange rate. [19]

See, for example, Dombusch and Fischer (1991), Andersen (1989) and Andersen (1992). [20]

This term is used in the sense of Gordon and King (1982) and others: the proportion of a year's GDP lost, or the percentage point rise in unemployment, per percentage point reduction in inflation. Some argue that, strictly speaking, these losses should be discounted to present value terms. In the present paper, since GDP losses have occurred mostly within the first year or two, discounting is ignored. [21]

See the discussion in Hagger (1978). [22]

Details of the HP filter are given in Blundell-Wignall et al. (1992). [23]

I am grateful to Malcolm Edey for suggesting this idea. [24]

References

Alogoskoufis, G. and A. Manning (1988), ‘Wage Setting and Unemployment Persistence in Europe, Japan and the USA’, European Economic Review, 32, pp. 698–706.

Andersen, P.S. (1989), ‘Inflation and Output: A Review of the Wage-Price Mechanism’, BIS Economic Papers No. 24.

Andersen, P.S. (1992), ‘OECD Country Experiences With Disinflation’, this volume.

Argy, V., A. Brennan and G. Stevens (1990), ‘Monetary Targeting: The International Experience’, The Economic Record, 66, pp. 37–62.

Argy, V. and J. Carmichael (1976), ‘Models of Imported Inflation for a Small Country – with Particular Reference to Australia’, in W. Kasper (ed.), International Money – Experiments and Experience, Papers and Proceedings of the Port Stephens Conference, Australian National University, Canberra.

Argy, V. and J.W. Nevile (eds.) (1985), Inflation and Unemployment: Theory, Experience and Policy Making, Allen and Unwin, London.

Arndt, H.W. (1957), ‘The Australian Economy, December 1957’, The Economic Record, 33, pp. 285–310.

Blanchard, O. and L. Summers (1986), ‘Hysteresis and the European Unemployment Problem’, NBER Macroeconomics Annual, pp. 15–78.

Blundell-Wignall, A. and R.G. Gregory (1990), ‘Exchange Rate Policy in Advanced Commodity Exporting Countries: The Case of Australia and New Zealand’, in V. Argy and P. de Grauwe (eds.), Choosing an Exchange Rate Regime: The Challenge for Smaller Industrial Economies, International Monetary Fund, Washington, DC.

Blundell-Wignall, A., P. Lowe and A. Tarditi (1992), ‘Inflation, Indicators and Monetary Policy’, this volume.

Blundell-Wignall, A. and G. Stevens (1992), ‘Fiscal Policy in Australia: Recent Developments and Current Issues’, in Current Issues in Fiscal Policy and Their Implications for the Conduct of Monetary Policy, Bank for International Settlements, Basle.

Boehm, E.A. (1984), ‘Money Wages, Consumer Prices, and Causality in Australia’, The Economic Record, 60, pp. 236–251.

Boehm, E.A. and V.L. Martin (1989), ‘An Investigation into the Major Causes of Australia's Recent Inflation and Some Policy Implications’, The Economic Record, 65, pp. 1–15.

Boxer, A.H. (1959), ‘The Australian Economy, November 1959’, The Economic Record, 35, pp. 299–322.

Butlin, M.W. (1977), ‘A Preliminary Annual Database’, Reserve Bank of Australia Research Discussion Paper 7701.

Carmichael, J. (1990), ‘Inflation: Performance and Policy’, in S.A. Grenville (ed.), The Australian Macro-Economy in the 1980s, Reserve Bank of Australia, Sydney.

Carmichael, J. and J. Broadbent (1980), ‘Inflation-Unemployment Trade-Offs in an Open Economy: Theory and Evidence for Australia’, Reserve Bank of Australia Research Discussion Paper 8010.

Caton, C.N., E.A. Evans and H.N. Johnston (1974), ‘The Inflation Process and the Impact of Wage Indexation on the Australian Economy’, paper presented to 4th Conference of Economists, Canberra.

Challen, D.W. and A.J. Hagger (1975), ‘Another Look at Australia's Inflation – Unemployment Trade-offs’, Australian Economic Papers, 14, pp. 137–153.

Coe, D.T. (1985), ‘Nominal Wages, The NAIRU and Wage Flexibility’, OECD Economic Studies, 5, pp. 87–126.

Cozier, B. and G. Wilkinson (1991), ‘Some Evidence of Hysteresis and the Costs of Disinflation in Canada’, Bank of Canada Technical Report No. 55.

Dawkins, P. and M. Wooden (1985), ‘Labour Utilization and Wage Inflation in Australia: An Empirical Examination’, The Economic Record, 61, pp. 516–521.

Dombusch, R. and S. Fischer (1991), ‘Moderate Inflation’, NBER Working Paper No. 3896.

Downing, R.I. (1956), ‘The Australian Economy, March 1956’, The Economic Record, 32, pp. 1–27.

EPAC (1991), Improving Australia's Inflation Performance, EPAC Council Paper, No. 48.

Firth, G.G. and A.J. Hagger (1959), ‘The Australian Economy, February 1959’, The Economic Record, 35, pp. 1–20.

Foster, R.A. and S.E. Stewart (1991), Australian Economic Statistics, Reserve Bank of Australia Occasional Paper No. 8.

Gordon, R.J. and S.R. King (1982), ‘The Output Cost of Disinflation in Traditional and Vector-Autoregressive Models’, Brookings Papers on Economic Activity, No. 1, pp. 205–244.

Gregory, R.G. (1986), ‘Wages Policy and Unemployment in Australia’, Economica, 53, pp. S53–S74.

Gregory, R.G. and R.E. Smith (1985), ‘Unemployment, Inflation and Job Creation Policies in Australia’, in V. Argy and J. Nevile (eds.), Inflation and Unemployment: Theory, Experience and Policy Making, Allen and Unwin, London.

Groenewold, N. and L. Taylor (1992), ‘Insider Power as a Source of Hysteresis in Unemployment: Tests with Australian Data’ The Economic Record, 68, pp. 37–62.

Grubb, D., R. Jackman and R. Layard (1983), ‘Wage Rigidity and Unemployment in OECD Countries’, European Economic Review, 21, pp. 11–39.

Gruen, F.H. (ed.) (1978), Surveys of Australian Economics, Allen and Unwin, Sydney.

Hagger, A.J. (1978), ‘Inflation’, in F.H. Gruen (ed.), Surveys of Australian Economics, Allen and Unwin, Sydney.

Higgins, C.I. (1973), ‘A Wage-Price Sector for a Quarterly Australian Model’, in A.A. Powell and R.A. Williams (eds.), Econometric Studies of Macro and Monetary Relations, North Holland, Amsterdam.

Hodrick, R.J. and E.C. Prescott (1990), ‘Post-War U.S. Business Cycles: An Empirical Investigation’, Carnegie Mellon University Discussion Paper 451.

Holmes, A.S. (1960), ‘The Australian Economy, February 1960’, The Economic Record, 36, pp. 179–190.

Jonson, P.D. (1973), ‘Our Current Inflationary Experience’, Australian Economic Review, 22, pp. 21–26.

Jonson, P.D. (1976), ‘Money Prices and Output: An Integrative Essay’, Kredit and Kapital, 9, pp. 499–578.

Jonson, P.D., K.L. Mahar and G.J. Thompson (1974), ‘Earnings and Award Wages in Australia’, Australian Economic Papers, 13, pp. 81–98.

Jonson, P.D., L.R. Moses and C.R. Wymer (1977), ‘A Minimal Model of the Australian Economy’ in W.E. Norton (ed.), Conference in Applied Economic Research, Reserve Bank of Australia, Sydney.

Karmel, P.H. (1961), ‘The Australian Economy, March 1961’, The Economic Record, 37, pp. 1–24.

Kasper, W. (ed.) (1976), International Money – Experiments and Experience, Papers and Proceedings of the Port Stephens Conference, Australian National University, Canberra.

Keating, M. (1982), ‘Australian Wage Determination’, Economic Papers, 1, pp. 64–73.

Kirby, M.G. (1981), ‘An Investigation of the Specification and Stability of the Australian Aggregate Wage Equation’, The Economic Record, 57, pp. 35–46.

Krause, L.B. and W.S. Salant (eds.) (1977), Worldwide Inflation, Brookings Institution, Washington, DC.

Layard, R. andC. Bean (1989), ‘Why Does Unemployment Persist?’, Scandinavian Journal of Economics, 91, pp. 371–396.

Lewis, P.E.T. and G.A. Macdonald (1991), ‘Cointegration Techniques and the Australian Wage Equation’, Economics Programme, Murdoch University Working Paper No. 52.

Liesner, T. (1985), Economic Statistics 1900–1983, The Economist Publications, London.

Lindback, A. and D.J. Snower(1987), ‘Union Activity, Unemployment Persistence and Wage-Employment Ratchets’, European Economic Review, 31, pp. 157–67.

Lowe, P. (1992), ‘The Term Structure of Interest Rates, Real Activity and Inflation’, Reserve Bank of Australia Research Discussion Paper 9204.

Lydall, H.F. (1962), ‘The Australian Economy, February 1962’, The Economic Record, 38, pp. 1–28.

Macfarlane, I.J. (1991), ‘The Lessons for Monetary Policy’, in I.J. Macfarlane (ed.), The Deregulation of Financial Intermediaries, Reserve Bank of Australia, Sydney.

Macfarlane, I.J. and W.J. Tease (1989), ‘International Interest Rate Linkages and Monetary Policy: The Case of Australia’, in International Interest Rate Linkages and Monetary Policy, Bank for International Settlements, Central Bank Economists' Meeting, 16–17 November 1988, Basle.

Matthews, R. (1958), ‘The Australian Economy, November 1958’, The Economic Record, 34, pp. 291–305.

McDonald, I.M. (1984), ‘Trying to Understand Stagflation’, Australian Economic Review, 3rd Quarter, pp. 32–61.

McDonald, J. (1975), ‘Wages and Prices in Australia: On the Short and Long-run Trade-offs Between Inflation and Unemployment’, Australian Economic Papers, 14, pp. 155–170.

McDonald, J. (1977), ‘The Relationship Between Wage Inflation and Excess Demand – New Estimates Using Optimal Extrapolative Wage Expectations’, The Economic Record, 53, pp. 490–507.

Meese, R.A. and K. Rogoff (1983), ‘Empirical Exchange Rate Models of the Seventies: Do They Fit Out of Sample?’, Journal of International Economics, 14, pp. 3–24.

Mitchell, W.F. (1989), ‘The Scandinavian Model of Inflation and Australian Wage Determination’, Flinders University of South Australia, Working Paper Series No. 102.

Nevile, J.W. (1977), ‘Domestic and Overseas Influences on Inflation in Australia’, Australian Economic Papers, 16, pp. 121–129.

Nieuwenhuysen, J.P. and J. Sloan (1978), ‘Wages Policy’, in F.H. Gruen (ed.), Surveys of Australian Economics, Allen and Unwin, Sydney.

Organisation for Economic Co-operation and Development (1973), ‘The International Transmission of Inflation’, in Economic Outlook, 13, pp. 81–96.

Parkin, M. (1973), ‘The Short-run and Long-run Trade-offs Between Inflation and Unemployment in Australia’, Australian Economic Papers, 12, pp. 127–144.

Parkin, M. (1976), ‘Yet Another Look at Australia's Short-run and Long-run Trade-offs Between Inflation and Unemployment’, Australian Economic Papers, 15, pp. 129–139.

Phipps, A.J. (1981), ‘The Impact of Wage Indexation on Wage Inflation in Australia: 1975(2)–1980(2) Australian Economic Papers, 20, pp. 333–350.

Pissarides, C.A. (1991), ‘Real Wages and Unemployment in Australia’, Economica, 58, pp. 35–55.

Pitchford, J.D. (1977), ‘Inflation in Australia’, in L.B. Krause and W.S. Salant (eds.), Worldwide Inflation, Brookings Institution, Washington, DC.

Poret, P. (1990), ‘The Puzzle of Wage Moderation in the 1980s’, OECD Department of Economics and Statistics Working Paper 87.

Rao, B.B. (1977), ‘An Analysis of the Short and Long-run Trade-offs Between Unemployment and Inflation and Estimates of the Equilibrium Steady State Unemployment Rate in Australia’, Australian Economic Papers, 16, pp. 273–284.

Reserve Bank of Australia, Annual Report, various.

Reserve Bank of Australia, Bulletin, various.

Richards, A. and G. Stevens (1987), ‘Estimating the Inflationary Effects of Depreciation’, Reserve Bank of Australia Research Discussion Paper 8713.

Rimmer, M. (1985), ‘Incomes Policy in Australia, 1975–1985’, University of NSW Working Paper No. 85-031.

Salant, W.S. (1977), ‘International Transmission of Inflation’, in L.B. Krause and W.S. Salant (eds.), Worldwide Inflation, Brookings Institution, Washington, DC.

Shrapnel, P.S., R.I. Dowing and J.O.N. Perkins (1958), ‘The Australian Economy, March 1958’, The Economic Record, 34, pp. 1–9.

Simes, R.M. and C.J. Richardson (1987), ‘Wage Determination in Australia’, The Economic Record, 63, pp. 144–161.

Simkin, C.G.F. (1972), ‘Inflation in Australia and New Zealand: 1953–71’, The Economic Record, 48 pp. 465–482.

Trivedi, P.K. and J. Payner (1978), ‘Wage Inertia and Comparison Effects in Australian Award Wage Determination, 1964–74’, The Economic Record, 54, pp. 195–218.

Waterman, A.M.C. (1972), Economic Fluctuations in Australia 1948 to 1964, Australian National University Press, Canberra.

Watts, M.J. and W.F. Mitchell (1990), ‘Australian Wage Inflation: Real Wage Resistance, Hysteresis and Incomes Policy: 1968(3)–1988(3)’, The Manchester School of Economics and Social Studies, 58, pp. 142–164.

White (1980), ‘A Heteroskedasticity-Consistent Covariance Matrix Estimator and a Direct Test for Heteroskedasticity’, Econometrica, 48, pp. 817–838.